CN103080984A - System and method for determining region-seen visibility and delta-PVS-based content stream processing using a conservative linearized umbral event surface - Google Patents

Abstract

A method determines a set of mesh polygons or fragments of the mesh polygons visible from a view region having a plurality of view region vertices, the mesh polygons forming polygon meshes. The method includes determining at least one supporting polygon between the view region and the polygon meshes. The method further includes constructing at least one wedge from the at least one supporting polygon, the at least one wedge extending away from the view region beyond at least the polygonal mesh. Further, the method includes determining one or more intersections of the wedges with the mesh polygons. Further, the method includes determining a set of said mesh polygons or fragments of said mesh polygons visible from the view region using the determined one or more intersections of the at least one wedge with said polygon meshes.

Description

Determine the observability of looking from the district and the system and method that carries out processing based on the content flow of increment PVS with conservative linearization umbra event face

Cross reference to related application

The application requires in the U.S. Provisional Application the 61/360th that is entitled as " System and Method of From-Region Visibility Determination and Delta-PVS Based Content Streaming Using Conservative Linearized Umbral Event Surfaces " of submission on June 30th, 2010, No. 283 the right of priority in the front applying date, the full content of this application merges among the application by reference.The application also requires in the U.S. Provisional Application the 61/382nd that is entitled as " System and Method of Delivering and Controlling Streaming Interactive Media Comprising Predetermined Packets of Geometric; Texture; Lighting and Other Data Which are Rendered on a Receiving Device " of submission on September 13rd, 2010, No. 056 the right of priority in the front applying date, the full content of this application merges among the application by reference.The application also requires the right of priority in the front applying date of No. the 61/384th, 284, the U.S. Provisional Application that is entitled as " System and Method of Recording and Using Clickable Advertisements Delivered as Streaming Interactive Media " submitted on September 19th, 2010.The full content of this application merges among the application by reference.The application also requires in the U.S. Provisional Application the 61/452nd that is entitled as " System and Method of Controlling Visibility-Based Geometry and Texture Streaming for Interactive Content Delivery " of submission on March 14th, 2011, No. 330 the right of priority in the front applying date, the full content of this application merges among the application by reference.The application also requires in the U.S. Provisional Application the 61/474th that is entitled as " System and Method of protecting game engine data formats and Visibility Event Codec Formats Employing an Application Programming Interface Between the Game Engine and the Codec " of submission on April 12nd, 2011, No. 491 the right of priority in the front applying date, the full content of this application merges among the application by reference.The application also requires the right of priority in the front applying date of No. the 61/476th, 819, the U.S. Provisional Application that is entitled as " System and Method of Delivering Targeted; Clickable; Opt-Out or Opt-in Advertising as a Unique; Visibility Event Stream for Games and Streaming Interactive Media " submitted on April 19th, 2011.The full content of this application merges among the application by reference.

Technical field

The present invention relates to a kind of method and system for come synthetic image with the visibility information of precomputation.

Background technology

In real time 3D figure viewing hardware has become and has become stronger day by day and afford.The availability of this hardware is so that computer and game console application program can show the scene that comprises several ten thousand graphic primitives in every frame routinely.Except a few exceptions, these hardware display systems adopt the hidden surface based on the Z buffer zone to remove algorithm.

Z buffer zone (Z-buffer) hidden surface removes the Z(degree of depth of the pixel (or segmentation) of algorithm each rasterizing (rasterized) by calculating each primitive in the view frustums) value solves the observability of every pixel.During rasterizing, the Z value of current segmentation compared with the existing Z value in the frame buffer zone and only in the Z of current segmentation value during less than the existing Z value in the Z buffer zone, the color of this segmentation is write frame buffer zone.

Although this method provides the acceptable performance for relatively simple scene, fails to put forward enough real-time performances for complicated, scene true to nature.Such scene often has high degree of depth complexity, and it forces usually, and each element to the Z buffer zone repeatedly compares during the playing up of single frame.In fact, be positioned at all hidden surface samples within the view frustums must be the Z rasterizing and it is compared to find immediate visible sample with the Z buffer values.

In some Z buffer zones are implemented, the Z that the rasterizing program is often not only carried out for all hiding segmentations determines to compare with the Z buffer zone, but also calculate to hide playing up fully of segmentation, only result's color is write frame buffer zone than close to existing Z buffer values the time when corresponding Z value.Even for the scene of moderate degree of depth complexity, this also can cause calculating waste and hydraulic performance decline.

Other Z buffer zones implement to comprise some types " early stage Z(early-Z) " refusal, if wherein the Z value of segmentation is greater than the Z buffer values of correspondence then disregard the color value of point counting section.This can reduce hiding playing up of segmentation, but only when play up with inferior ordered pair graphic primitive from back to front efficient reach maximum.

Another modification to hardware Z buffer zone is " level Z buffer zone " algorithm (people 1993 such as Greenet) (Green N., Kass, M., Miller, G " hierarchical Z-Buffer Visibility " Proceedings of ACM Siggraph1993pp.231-238, its full content merges among the application by reference) some key element integrated.This algorithm adopts the layering of Z buffer zone to carry out fast observability refusal test.Although realized the basic version of level Z buffer zone pyramid itself in some system (for example Nvidia, ATI), level Z buffer zone algorithm has been proved to be and has been difficult to implement at hardware completely.In these were implemented, the low-definition version of Z buffer zone was kept in the storer of each rasterizing program element for this locality.Use the expression of these this locality in " early stage Z " refusal test of formerly describing.If can compare to refuse each segmentation by the Z buffer zone element that each segmentation and this locality of low resolution are stored, thereby avoid the slower access to high resolving power (non-local) Z buffer zone.

In the hardware Z of these acceleration Buffer System, " early stage Z " refusal sometimes can stop to the segmentation of hiding play up and level Z pyramid can be accelerated the test of " early stage Z " refusal.Yet such accelerating system still requires: by how much stages of graphics pipeline process in the view frustums all primitives and by Z at least generate/the refusal test processes all segmentations that include the face of being blocked in the view frustums mutually.Therefore, when the scene of high degree of depth complexity was played up, these systems still showed bad luck.

In view of the relatively relatively poor performance for the Z Buffer System of the scene of high degree of depth complexity, developed following algorithm: the geometric figure that identification is blocked and get rid of so geometric algorithm the stage from how much stages and the rasterizing of hardware graphics streamline.These occlusion culling technology can be carried out in working time or pretreatment stage.The summary of visibility culling technology is published in in the Publication about Document: the people such as Cohen-Or (2003) (Cohen-Or, Daniel, Deng people " A Survey of Visibility for Walkthrough Applications. " IEEE Transactions on Visualization and Computer Graphics9.3 (2003): 412-31.Print., its full content merges among the application by reference.Visibility culling refers to remove any method that (namely by the Z buffer zone) identified and refused sightless geometric figure before at the hidden surface that carries out reality.With graphic primitive from the application that hardware pipeline is rejected, usually adopt the back side of the use level technology of having established well to reject and the method for view frustums rejecting.Occlusion culling is to avoid one type visibility culling method that the primitive that is blocked in the scene is played up.Occlusion culling relates to the mutual relationship of the complexity between the graphic primitive in the model, and compare view frustums reject common carried out more difficult.

Usually, working time, the occlusion culling technology determined which type of geometric figure is to look visible from single viewpoint.The rejecting technology that these technology is called " looking (from-point) from point ".On the contrary, determine any viewpoint from designation area geometric subset of looking visible for the preprocess method of occlusion culling.Latter's method is called the observability technology of " (from-region) looks from the district ".

" roaming " type application is paid close attention in the people's such as Cohen-Or (2003) investigation, and this class is used and is characterised in that relatively a large amount of static geometric figures and highly possible degree of depth complicacy.Many computer games, simulator and other interactive application all belong to this class.When being applied to the occlusion culling technology of " looking from the district " the geometry data storehouse in pre-treatment step, these application are often greatly benefited.These technology become district or unit with model partition.But these look the cruising ground that can comprise viewpoint that the unit is model.During pre-processing, determine graphic primitive from looking any place in the unit potentially visible subset (potential visible collection or PVS) of looking.The major advantage of the observability technology of looking from the district is pre-treatment step rather than pays sizable occlusion culling in working time and assess the cost.

The observability preconditioning technique of usually, looking from the district is intended to calculate conservative the over-evaluating for the accurate PVS that looks the unit.The first observability method of looking from the district is to observe and develop for the interactivity of BUILDINGS MODELS.The observability that BUILDINGS MODELS is subdivided between unit (for example, room, entrance hall) and the unit naturally manifests by the connector (doorway, window) that is called as entrance.Airey(1990) to have developed this structure in simple, the axially aligned model.He has proved following methods: that usefulness is similar to but conservative shade umbra calculates to identify by the visible polygon of entrance.

Teller(1992) (.Teller, Seth, Visibility Computations in Densely Occluded Polyhedral Environments.Diss.U of California at Berkeley, 1992.Berkeley:U of California at Berkeley, 1992.GAX93-30757.ACM Portal, its full content merges among the application by reference) and Sequine the method for the observability of looking from the district of unit-entrance expand to do not require that the user limits the polygon model of non axial aligning of wall and entrance.The BSP tree (automatically division) that Teller adopts the polygon by model to limit.The leaf of BSP tree must be convex polygon, and it may be fully not closed.These convex polygons are the observability unit (or looking the unit) of model.Use available cell abutment information in BSP figure is identified the borderline open region between the adjacent unit and it is enumerated as entrance between the observability unit.

Therefore, the visibility problem that will look from the district by a series of polygon entrance of the Teller structure of utilizing BSP automatically to divide is reduced to more limited and the visibility problem of simplifying.Teller points out: even for relatively limited visibility problem, the visible volume that will look from the unit and look from the unit be blocked volume integral from the visibility events face also be quadric surface usually.

Teller is by adopting the test for the existence of the extreme value ruler that passes entrance or entrance sequence between the unit to come determining unit to the observability of unit.In this method, come determining unit to the observability of unit by confirm existing to come from source unit and to penetrate a series of entrances at least one ray of linkage unit.For example, pass through the existence that ray provides the such ray that passes four entrances by the extreme value of incident on any four limits of relevant entrance.Use is identified such ray with Pu Lvke (Plucker) mapping that the line in the three dimensions is mapped to the plane in the quintuple space.The cross-shaped portion on these four planes forms line in quintuple space, and this line and Pu Lvke quadric surface intersect and generation two non-void (non-imaginary) result at the most.Each cross-shaped portion in these cross-shaped portions is corresponding to the line that intersects with four entrance limits in three dimensions, i.e. extreme value ruler.The cost that the location extreme value is passed through ray is the O(n of the quantity on the limit in the entrance sequence ²).Because increment ground is carried out and passed through, total cost is O(n ³).The method adopts the singular value matrix decomposition method, because the geometric degeneration that runs in crossing sequence, it can show numerical instability.

Teller has also developed a kind of method of calculating accurate visible volume by the entrance sequence: non-penumbra (antipenumbra) volume.Record as previous, in the ordinary course of things, this volume is surrounded by plane and quadric surface.In the method, by the dual Plucker coordinates that changes into, every line in 3 dimension spaces represents the coordinate on the plane in 5 dimension spaces again on the limit of entrance.With more high-dimensional convex hull computation the plane corresponding with all limits in the entrance sequence intersected each other to form polyhedron in 5 dimension spaces.This polyhedral and the quadric cross-shaped portion of Pu Lvke are corresponding to extreme value band (swath) or visibility events face between the entrance limit.Directly do not calculate 5D face and Pu Lvke quadric crossing, intersect but the calculating polyhedral limit of 5D and Pu Lvke are quadric.The polyhedral limit of 5D and Pu Lvke are quadric to intersect corresponding to the extreme value ruler that surrounds band.These 5D limits and the quadric cross-shaped portion of Pu Lvke are to identify by the root of seeking quadratic equation.These bands are by calculating 5D limit and the quadric cross-shaped portion of Pu Lvke and checking that the polyhedral face (limit among the 3D) of sharing this 5D limit of 5D comes non-directly identification.

Each band can be the composition on the border of non-penumbra, or alternatively, can be positioned at non-penumbra volume fully.Comprise (containment) test and be used for the identification boundary strip.

Teller finds to be difficult to robust ground and implements non-penumbra calculating.The method needs the linear programming of high latitude to calculate and the root finding method, its together fully robust ground for complex model.

Teller(1992) and Teller and Hanrahan(1993) (Teller, Seth J., and Pat Hanrahan. " Global Visibility Algorithms for Illumination Computations. " Proceedings of the20th Annual Conference on Computer Graphics and Interactive Techniques.New York:ACM, 1993, its full content merges among the application by reference) also developed a kind of for determining unit to unit and unit the better simply technology of observability of determining to pass the entrance sequence to object.In this is implemented, conservatively be similar to non-penumbra by convex polyhedron.Be somebody's turn to do " linearization " non-penumbra by the separating plane encirclement of the convex closure that effectively forms this non-penumbra of entrance sequence.The plane that defines the border of the non-penumbra of linearization intersect each other and with the encirclement Plane intersects of BSP leaf unit to determine to pass the observability of entrance sequence.

Although non-half image method of linearization has been over-evaluated the unit that passes the entrance sequence to the observability of unit, it is applicable to robust and implements.

In 1996, John Carmack used a kind of for precomputation for the unit of the computer game Quake method to the observability of unit.The observability pre-computation methods in Quake of Carmack is a bit similar to non-half image method of the described linearization of Teller.In the method for the method of Teller and Carmack, by large shelter (for example, wall, floor) is set as the BSP of splitting plane the geometry data storehouse is segmented.The terminal leaf of this segmentation is the convex polyhedron that possible have one or more non-closed borders or entrance.In these two kinds of methods, the entrance between the leaf unit is identified and used the linearization of the non-penumbra between the entrance over-evaluated and come determining unit to the observability of unit.

In the method for Teller, construct linearizing non-penumbra by the rotation on the summit two specific extreme values or " separation " from each entrance limit to the entrance sequence: each semispace that is arranged in entrance.(extreme vertex of entrance is the summit that forms separating plane with original entrance limit between two entrances.) selected extreme vertex is created in the plane with entrance and every other extreme vertex is arranged in identical semispace.

In the enforcement of Carmack, the paired order of the non-penumbra of linearization intersects the existence of the observability of the unit-unit that is used for definite entrance chain.The reality of not carrying out the object in non-penumbra and each unit intersects.Be for the unit of each the leaf unit PVS to the unit with result store.

Carmack passed use to the enforcement of Teller algorithm entrance in 1996 unit segments the method for optimizing that is asserted for the observability precomputation of computer game to the BSP space of the observability of unit.The 3D computer game system of directly deriving from the Quake code of Carmack afterwards (for example, " Source " game engine of Quake II, Quake III and Valve Software) or with its irrelevant 3D computer game (for example, " Unreal " game engine of Epic Games company limited) adopted the precomputation occlusion culling method for the polyhedron environment that blocks thick and fast.

In these all systems, " outpost of the tax office editor " instrument of use creates geometric configuration and the modeling environment other static state, that potential environmental element of blocking is played with structure of wall, floor and ceiling.Then this geometric figure is committed to the pre-service of constructing the BSP tree with traditional B SP algorithm according to this geometric configuration.Then, usually calling the second pre-service to calculate the unit of each leaf unit of setting for BSP to the PVS of unit with previously described observability method of passing entrance.Usually will be stored as for the PVS of concrete leaf unit effective compressed-bit vector, its expression is from source unit other BSP leaf unit of looking visible.

Show that in working time the concrete leaf comprise current view point between haplophase, determines viewpoint leaf unit with simple BSP algorithm.Read the PVS for viewpoint leaf unit, then the level of Application standard is looked (potential visible) leaf unit that the centrum elimination method rejects about working as forward sight centrum level ground correspondence.Then will send to the hardware graphics streamline from those graphic primitives of the PVS leaf unit that is positioned at view frustums.During show working time, can also for example further limit the primitive that sends to the hardware graphics streamline from entrance and the rejecting of non-entrance that point is looked with various occlusion culling methods of looking from point.Yet the PVS of precomputation normally carries out the working set of the primitive of the rejecting of looking from some working time to it.Thereby, precomputation PVS to working time performance very important, this is not only because paid its own occlusion culling cost in pre-service, but also can reduce by limiting the geometric amount that must operate on it the cost of the occlusion culling method of looking from some working time because of PVS accurately.

Although the BSP/ entrance-sequence method of PVS precomputation is widely used for strengthening counting game and the similar performance of using, the current enforcement of the method still has many shortcomings.As previously discussed, to the linearization of the non-penumbra of entrance sequence is approximate can be so that the method be over-evaluated the size of PVS significantly.

Another restriction of this method is: need to construct BSP(according to the geometric figure that blocks potentially of model and automatically divide).Known with the space segmentation of setting to carry out with the effective BSP in space of equilibrium is hard problem (seeing 96 pages of TELLER(1992) inherently).For the size of the tree under worst case O(n ²), the best boundary of the time complexity of tree structure is O(n ³).In the situation that select well, can produce the heuristic BSP of the division with fair-sized for the model of moderate complexity.But for larger model, these times and space cost function can be so that forbid actual BSP structure and storage.Thereby when using the method, the user must often limit the quantity of the primitive that is used for structure BSP.The complex object that comprises a large amount of non-coplanar primitives is excluded as potential shelter usually, and this is because they can increase time and the space cost of BSP structure.Usually come by the following method to manage individually such object: the method needs user (that is, outpost of the tax office deviser) above-mentioned object to be appointed as " details " object that the BSP plane is not provided and is not used as shelter during the PVS precomputation.In the method still can be with these " details " objects as the potential thing that is blocked.If the details object is included in the PVS leaf unit fully and the leaf unit is not confirmed as for the unit of the given viewpoint leaf unit part to the PVS of unit, then the PVS of details object from viewpoint leaf unit can be got rid of.Although substitute their possibility of blocking by object-based geometric complexity object is got rid of from the consideration as potential shelter, the method can be over-evaluated the actual PVS that looks from the district significantly.

The relevant weakness of BSP/ entrance-sequence method is: for the modeling environment except building interior, and the method poor performance.When being applied to the building interior model, the method is easy to naturally to construct and has the BSP leaf unit corresponding with the room of entrance corresponding to door or window.By contrast, for the inside scene of opening, Outdoor Scene and many complexity, observability is concerned to arrange by closed cell, opening-entrance more unintelligiblely.In these scenes, the independently shelter that observability often mainly is subjected to not to be associated with the unit of relative closure limits.When geometric figure that rejecting is blocked, BSP/ entrance-sequence is not considered each independently fusion of shelter effectively.BSP/ entrance-sequence method is applied to such scene can produces very large BSP and very long entrance sequence.Under these conditions, the method trends towards spending very long time and calculates the PVS low with run-time efficiency that has been over-evaluated to heavens.Use the application of BSP/ entrance-sequence method usually can avoid precomputation for the PVS precomputation of such scene and alternatively rely on the entrance rejecting of occlusion culling method as looking from point of looking from point; for example, by the Source of the Valve Software that must calculate in working time ^RThe employed dynamic non-entrance method of game engine.

Teller comprises following technology to the initial description of entrance sequence method: by coming computing unit to the PVS of primitive each primitive intersection in the non-penumbra of linearization and the BSP leaf unit.In practice, this technology is adopted by Carmack or other existing systems, partly because the carrying cost of the PVS of unit-primitive can be high more a lot of than the cost of the PVS of unit-unit.

Although used the various BSP/ entrance sequence observability pre-service that are similar to simplify and accelerate, it remains and calculates expensive processing.Because BSP/ entrance sequence method is over-evaluated PVS, so the graphic primitive that fully is blocked may experience expensive pre-service working time, although its graphic primitive that fully is blocked is sightless in scene.Can be paid by CPU, GPU or both the assessing the cost of primitive that runtime processing is blocked.CPU processes the CPU cost can comprise that non-entrance that the entrance that view frustums is rejected, looked from point is rejected, looked from point is rejected and primitive submitted to GPU in batches.On the GPU side, the processing stage of summit that the primitive that is blocked may experience the hardware graphics streamline and the rasterizing stage.A kind of measurement of the efficient of precomputation occlusion culling is the degree of the overdraw that occurs in working time.In the time the non-NULL clauses and subclauses in rasterizing segmentation and the Z buffer zone must being compared, overdraw just might appear during rasterizing.Non-NULL clauses and subclauses in the Z buffer zone are that the rasterizing early by the segmentation that is in same image space coordinate place causes.Clauses and subclauses early can be positioned at front or back (being blocked) of current segmentation.This situation must read with comparison operation by the Z buffer zone and solve.If the Z value of clauses and subclauses early is more farther than the Z value of current segmentation, then override clauses and subclauses early.As described earlier, modern hardware Z Buffer System can use " early stage Z " refusal test that can comprise level Z comparison mechanism prevent from being blocked painted (shading) of reality of segmentation sometimes.Yet, make it compare in the rasterizing segmentation that has at least them each with corresponding Z buffer zone and/or its level equivalent to the primitive that fully is blocked in the rasterizing stage of graphics pipeline.We adopt following convention: overdraw comprises any " overlapping " of the segmentation in the image space that needs at least the Z comparison operation.

When the BUILDINGS MODELS of the Quake that BSP/ entrance sequence method is applied to play, find that but average overdraw is 50% can be changed to 150% in the worst case.(Abrash1997, pg.1189, Abrash, Michael " Michael Abrash ' s Graphics Programming Black Book Special Edition ", 1997The Corilois Group, its full content merges among the application by reference).The overdraw of this level can run into for simple model relatively the time, and it has and is approximately 10 depth capacity complexity and often minimizes wittingly the visible depth complexity by the position of carefully selecting to block wall or entrance therein.

In the Quake of ID Software III computer game, used rear a kind of enforcement of the observability pre-computation methods of Carmack.Compare original Quake game, the environment of simulation has obviously more geometric figure details (about 40,000 polygons in every outpost of the tax office) in this game.As in original game, carefully design the outpost of the tax office to comprise various barriers, it comprises the wall of passageway, right angle, back, doorway, the stair with u turn and other observability restraining masss.Therefore these barriers are arranged to the observability in the limited model wittingly, have reduced the size for the PVS of the observability unit of model.Even have these observability restraining masss, the entrance observability of approximate unit-unit is calculated during show working time or can be caused sizable overdraw.When BSP/ entrance sequence pre-computation methods was applied to Quake III, the general generation of BSP/ entrance sequence pre-computation methods was typically 80% overdraw and surpasses 300% in the worst case.These results obtain in the complexity that typically fathoms during the roaming working time at the Quake III outpost of the tax office by usefulness-dc command-line option.During these are measured, must control carefully for the painted effect of multiple tracks.

Even because when BSP/ entrance sequence method is applied in its only modeling environment, the method also is to calculate expensive and inefficient occlusion culling method of looking from the district relatively.Thereby more how close next work concentrates on the occlusion culling method of looking from the district that can be applied to general scene and produce more accurate PVS reasonably to assess the cost.

The method of the early stage general occlusion culling of looking from the district of guarding has been described: the people .(1998 such as Cohen-Or) (Chrysanthou in Publication about Document, Yiorgos, Daniel Cohen-Or, and Dani Lischinski. " Fast Approximate Quantitative Visibility for Complex Scenes. " Proceedings of the Computer Graphics International1998.Washington, DC:IEEE Computer Society, 1998.220 its full content merges among the application by reference).In these methods, only have when object is blocked by single, large, protruding shelter just with its rejecting of object.Regrettably seldom running in actual applications large protruding shelter exists.

The observability pre-computation methods of looking from the district of (shelter fusion) is blocked in the combination of more recently, having developed the collection of attempting considering less shelter.

The people such as Durand (2000) (Durand, Fredo, et al. " Conservative VisibilityPreprocessing using Extended Projections. " Proceedings of the27th Annual Conference on Computer Graphics and Interactive Techniques.Proc.of International Conference on Computer Graphics and Interactive Techniques.New York:ACM Press/Wesley Publishing Co., 2000.239-48 its full content merges among the application by reference) a kind of observability pre-computation methods of looking from the district of the conservative image-space representation that is called as generalized projection of using shelter and the thing that is blocked proposed.In the method, by the shelter primitive of looking from the viewpoint different from corresponding eight in the summit of looking the unit is carried out the conservative expression based on pixel that rasterizing is constructed protruding shelter.The generalized projection of protruding shelter is that it is from the cross-shaped portion of the projection of these viewpoints.Can be by the shelter rasterizing be calculated this cross-shaped portion in the hardware Z buffer zone that forms together " broad sense depth buffer " and stencil buffer data structure.Be blocked thing by the bounding box that conservatively is expressed as them from the identical associating of looking the projection on summit, unit.The shelter that thing is positioned in the broad sense depth buffer if be blocked covers fully, then since the above-mentioned thing that is blocked look from this district invisible, so its above-mentioned shelter is rejected.The generalized projection of a plurality of shelters is aggregated in the broad sense depth buffer, and it has explained the shelter fusion.

The method can use with around broad sense depth buffer corresponding to the singleton on six planes of whole environment.Alternately, can use apart from the continuous collection around the plane of the ever-increasing distance of viewpoint unit.In this case, can use conservative convolution operator will project at the shelter of the polymerization on the hither plane on subsequently the plane.Should " block and scan " re-projection method catching apart from more effective aspect the fusion of a plurality of little shelter of the change distance of viewpoint unit.For example, this arranges the shelter polymerization that is used for explaining at the scale Forest Scene of high degree of depth complexity.

This generalized projection method has been used a large amount of being similar to of over-evaluating that causes PVS.At first, because the method does not have to use the projection of the thing self that is blocked, so the size of the potential thing that is blocked is always over-evaluated.Alternatively, the projection bounding box of thing that is blocked.In addition, second of this projection the area-encasing rectangle that is similar to is for the generalized projection of calculating the thing that is blocked.These continuous approximations cause the over-evaluating of the size of the thing that is blocked, and have therefore reduced the precision of PVS.In addition, effectively limited the precision of generation unit to the method for the PVS of object (rather than the unit is to primitive) for the be blocked requirement of thing bounding box of use.

The generalized projection method can be directly only with the depth buffer of protruding shelter rasterizing to broad sense.At first must recessed shelter be converted to protruding expression by recessed shelter and projection plane are intersected.This is an additional step that needs object-space to calculate, and depends on the feature of the object plane that is blocked, and this step may be to calculate the ground costliness.In addition, if the position of projection plane is undesirable, crossing calculating can be underestimated the actual occlusion effect of recessed shelter significantly.

It is with the technology of shelter from a projection plane re-projection to a farther projection plane that the generalized projection method is used another kind of being similar to.The target of this re-projection is effectively to identify the umbra of plane shelter (with respect to by the light source of looking the unit and representing) and find this umbra and the intersecting of farther plane.The generalized projection method estimates conservatively that by the image of shelter and inverse video as the rectangle of over-evaluating of the light source that is formed by the viewpoint unit are carried out convolution this is crossing.This technology can be underestimated the umbra of similar to the viewpoint unit in size shelter significantly.The size of the shelter by underestimating significantly re-projection, the method can be easy to over-evaluate PVS.

The mainspring of generalized projection method is to detect blocking that combined effect by a plurality of little shelters causes.The people such as Durand (2000) admit that the method only detects the fusion between the crossing shelter of the umbra of shelter (district that is blocked), and when intersecting volume self and selecting arbitrarily the parallel projection Plane intersects.Because used relatively few projection plane in the enforcement blocking to scan, the method often fails to detect the fusion of blocking that the umbra that intersected by near projection plane outside causes.

The people such as Schaufler (2000) (Schaufler; Gernot; et al. " Conservative Volumetric Visibility with Occluder Fusion. " Proceedings of the27th Annual Conference on Computer Graphics and Interactive Techniques.New York:ACM Press/Wesley Publishing Co.; 2000.229-38; its full content merges among the application by reference) developed the conservative PVS that looks from the district of a kind of precomputation, it needs the volume of modeling environment to represent.In the method, modeling object must be surrounded by closing face.Closed enveloping surface must produce the internal volume that defines well for each object.The internal volume of supposing object is that convex body opaque and that generated by the volume decomposition by inside usually represents.Voxel is as the shelter for the method.Calculate by the axle (shaft) that seek to connect viewpoint unit and voxel and to block.The extension of this axle is the umbra that all within it geometric figures are blocked.The method explains that by the bordering compounding voxel and by the adjacent region in combination voxel and the space that is blocked shelter merges.The enforcement that proposes for the PVS(of the unit-unit of 2D and 2.5D environment is for example calculated, and is modeled as the city of height field).Although the author discusses the expansion of full 3D environment, the calculating that the details volume of 3D model represents and carrying cost are the real restrictions of the method.Although it is protruding that the people's such as Schaufler volume observability method does not need shelter, needing shelter is the stream shape with discernible solid (waterproof) internal volume that has formed well.This makes it possible to conservatively be similar to each shelter by the box-like structure that is positioned at original shelter inside fully.Should approximate shelter be by in the processing of the restraining mass expansion of attempting the approximate shelter of the included box-like of maximization, voxel being resolved in inside and these voxels of recombinating generate.The shelter that the method need to be similar to keeps the box-like shape to be conducive to the structure be used to the axle of determining to block.The major limitation of the method is that the bad ground of many shelters is similar to by included box-like structure.Particularly, recessed object or object (deficiency greater than 0 stream shape) with topological hole present for the unclean situation of restraining mass extended algorithm and have underestimated significantly by blocking that object causes.Main is that the 2.5D enforcement of method that is used for looking the PVS of unit in 0 the figure object calculating city model coming the people such as Schaufler are described is tested with deficiency.Use the inner shelter of box-like to be easy to be similar to these objects pretty goodly.For the more real model that comprises recessed element and hole (for example, door and window), the method is lower to efficient on shelter approximate, and therefore efficient is also lower on to the geometric rejecting that is blocked.

Shelter in the situation that the linearization umbra of this volume observability method detection shelter intersects merges.But for independent shelter, the restraining mass extended algorithm has finally produced the box-like of the simplification of the polymeric area that blocks approximate, and it can underestimate the effect that shelter merges significantly.

Generalized projection method and volume observability method will be looked effectively the unit is processed as area source and be calculated the conservative linearization of the umbra of polygonal mesh approximate with image-space and object-space technology respectively.The algorithm that is used for the shadow edge (umbra and penumbra) of calculating polygon facet light source, Nishita, Nakame(1985) (Nishita, Tomoyuki, Isao Okamura, and Eihachiro Nakamae. " Shading Models for Point and Linear Sources. " ACM Transactions on Graphics (TOG) 4.2 (1985): 124-46, its full content merges among the application by reference) and Chin-Finer(1992) (Chin, Norman, and Steven Feiner. " Fast Object-Precision Shadow Generation for Area Light Sources Using BSP Trees. " Proceedings of the1992Symposium on Interactive3D Graphics.Proc.of Symposium on Interactive3D Graphics, 1992, Cambridge, Massachusetts.New York:Association for Computing Machinery, 1992, its full content merges among the application by reference) also utilized conservative linearization umbra border.

These shadow edge methods have only been used the linear umbra event face that is formed between single convex polygon light source and the single convex polygon.For example non-convex polygon grid is used these methods to produce can accurately not represent the discontinuous umbra event face of umbra volume.Thereby in fact the use of these methods is restricted to very simple model.

In 1992, Heckbert(Heckbert, P " Discontinuity Meshing for Radiosity ", Third Eurographics Workshop on Rendering, Bristol, UK, May1992, pp203-216, its full content merges among the application by reference) used and be called incomplete uncontinuity gridding and construct from the accurate linear visibility events face (umbra and penumbra) of area source by the projection of simple polygon model.In this technology, be formed with linear event face or wedge between the summit of the limit of light source and shelter and between the limit of the summit of light source and shelter.Wedge and all model polygon intersect, and polygonal in fact on wedge visible segmentation follow-up by using the observability algorithm (Weiler that looks from point in Weiler-Atherton object-space, Kevin, and Peter Atherton. " Hidden Surface Removal using Polygon Area Sorting. " Proceedings of the4th Annual Conference on Computer Graphics and Interactive Techniques.New York:ACM, 1977.214-22 its full content merges among the application by reference) the 2D version determine.

The mainspring of uncontinuity gridding method is the uncontinuity border in the identification penumbra.These borders can be for increasing the precision of the brightness calculation in the penumbra.Regrettably because incomplete uncontinuity gridding method has only been constructed accurate linear umbra event wedge, so it generally fail to produce completely, continuous umbra event face.This is because for all models except the simplest model, and continuous umbra event face (for example, incident on the profile contour of polygonal mesh) is that visibility events face and the secondary visibility events face by the plane forms.Thereby incomplete uncontinuity gridding method is unsuitable for identification and looks as seen or the grid polygon or the segmentation of grid polygon that are blocked from area source (or looking the unit).

In the incomplete uncontinuity gridding method of prior art, all visibility events faces all are to be formed by summit and limit.Figure 53 is the incomplete uncontinuity gridding method by the Heckbert proposition from prior art.The figure shows accurate linear visibility events face or wedge, such as the triangular structure wedge R of crested ".Be labeled as wedge R " wedge incident and in the above-mentioned incident of vertex v that can be the summit of light source on polygonal limit e.In incomplete uncontinuity gridding method, on linear event face is not limited to and looks the segmentation on sightless limit from the summit.In the situation of Figure 53, " wedge R " is not limited in the segmentation that is labeled as GAP E of limit e, because look from GAP E, polygon O has blocked vertex v.Because wedge is not limited in this segmentation, so wedge and polygon P crossing caused gap corresponding between " segmentation 1 " and " segmentation 2 ".If wedge R is the umbra wedge, then its phase intersection with P produces incomplete umbra border.Because these gaps, so can not use separately the linear visibility events wedge by incomplete uncontinuity gridding method structure to limit umbra border (or the observability border of looking from the district).

Drettakis and Fiume(1994) (Drettakis, George, and Eugene Fiume. " A Fast Shadow Algorithm for Area Light Sources Using Backprojection. " Proceedings of the21st Annual Conference on Computer Graphics and Interactive Techniques.New York:ACM, 1994.223-30 its full content merges among the application by reference) fully characterized the visibility events face between the object that appears in polygon light source and the polyhedron environment.In being called as completely the uncontinuity gridding method, umbra event face and penumbra event face are identified and intersect with the model polygon.These intersect the model geometric figure is divided into " completely uncontinuity grid " so that the view of light source is equal in each face on topologys.This uncontinuity grid is illustrated as for the useful data structure of calculating the overall brightness in the penumbra.

In uncontinuity gridding method completely, four class event faces between area source (being called " luminophor ") and the polyhedron grid object are identified.It is quadratic form that two of these event faces are the plane and other two kinds.

The first kind visibility events face of identifying is formed between the certain edges thereof or summit of the summit of luminophor or limit and polyhedral model.These polygons are called luminophor-VE or (E-EV) wedge.The author emphasizes that all summits that are not the polyhedron grid all support the E-EV wedge.Only be those Grid Edges of the silhouette edge (they are called " shade limit ") of looking from point support wedge just for the arbitrfary point on the luminophor.Identify all Grid Edges that support umbra or penumbra E-EV wedge by limiting by this way " silhouette edge of looking from the district ".

The visibility events face that is used in the another kind of plane in the completely uncontinuity gridding is the wedge of non-luminophor-EV(NonE-EV).This type of wedge is formed between any limit of polyhedron grid and any other limit potentially so that formed wedge and luminophor intersect.For any limit of polyhedron grid, the NonE-EV wedge that supports only appears in the axle that is formed between this limit and the luminophor.This true conformation identification NonE-EV wedge that is used for.

The 3rd class visibility events face is according to limit of luminophor and two quadric surfaces that the limit forms of polyhedron grid.This is called as luminophor-EEE event face or EeEEE face.Whenever the place that two of the uncontinuity grid non-adjacent hypotenuses intersect is just identified this.[this intersect in fact with the formation compound-contoured contour of plane wedge and the silhouette edge of looking from the district crossing corresponding].The continuous visibility events face that is in this point is quadric surface.

The 4th class that forms between planar luminous body and polyhedron grid object and last class visibility events face are NonE-EEE.This is formed between three hypotenuses of polyhedron grid so that result's quadric surface and the secondary event face of looking the unit and intersecting.

In this manual, some of nomenclature revise to adopt based on Drettakis and Fuime(1994) the classification of the visibility events face of looking from the district to adapt to further subclassification.Table I a comprises former cause Drettakis and Fuime(1994) the four class visibility events faces that propose, for the sake of clarity, it is renamed.

Table I a

The prior art nomenclature of the visibility events face of looking from the district

In the four class visibility events faces any one can finally be facilitated actual in the look umbra border of (looking from the district) of luminophor, and it will be from the luminophor, and the spatial volume that is blocked and the arbitrfary point from luminophor the spatial volume of looking visible of looking a little separates.Regrettably, use existing uncontinuity grid method, do not have existing method to be used for determining which event face can facilitate this umbra border that the observability of looking from the district is limited.Thereby, in order to identify umbra visibility events conservative, that look from district border with the uncontinuity gridding method, at first must generate all visibility events faces and must carry out to result's uncontinuity grid aftertreatment and represent the actual umbra border of looking from the district to determine which event face-grid polygon cross-shaped portion.

Some other problemses are restricted to the conservative observability of looking from the district of calculating with the use of uncontinuity gridding method.The secondary event face is so that be difficult to carry out the projection of event face.Need the event projection surface to find the visible visible quafric curve section (in the situation that luminophor EEE wedge) of looking from the luminophor limit.The 2D that observability typically uses Weiler-Atherton observability algorithm on this wedge implements to solve, and above-mentioned algorithm is being difficult to the enforcement of robust ground when using quadric surface.

As discussed previously, if dispense simply quadric surface (as in incomplete uncontinuity gridding method), then can not guarantee the continuous umbra face of looking from the district and so that can not determine the visible grid polygon of looking from the district.

Another critical limitation of traditional uncontinuity gridding method is that it does not demonstrate the responsive performance of output.This is because existing uncontinuity GRIDDING WITH WEIGHTED AVERAGE comes with all the visibility events faces on the silhouette edge of all (looking from the district) of generating the polyhedron grid.This comprises the silhouette edge of looking and in fact being blocked from luminophor/source.These event faces are then crossing with each potential polygon of polyhedron grid, and follow-up as aftertreatment, use 2D Weiler-Atherton observability that visible segmentation on the wedge is identified.Because any stage at these algorithms does not all have depth-first, so they block thick and fast carry out in the environment very bad, wherein, the most of border that generates can be positioned at the conservative umbra border of looking from the district, therefore is helpless to the observability solution of looking from the district.

As shown subsequently in this manual, this method of utilizing the conservative linearization umbra event face that generates with the responsive algorithm of output to construct observability figure has solved the restriction of existing uncontinuity gridding method when the visibility problem of looking from the district that is applied to guarding.

Use is by Drettakis and Fiume(1994) during the classification of the visibility events face of looking from the district described, be apparent that, volume observability method (Schauffler2000) has only utilized the E-EV face to represent the umbra border.Generalized projection method (and other projection methods) has also implicitly been used E-EV umbra border.

The image of many conservative observability precomputations of looking from the district-space technology uses " shelter of contraction " conservatively to be similar to the observability of looking from this district with the observability of using a single point from the district looks.The people's such as Wonka (2000) method (Wonka, Peter, Michael Wimmer, and Dieter Schmalstieg. " Visibility Preprocessing with Occluder Fusion for Urban Walkthroughs. " Proceedings of the Eurographics Workshop on Rendering Techniques2000.London:Springer-Verlag, 2000.71-82. its full content merges among the application by reference) use the method conservatively to calculate the observability of looking from the district around the viewpoint that is positioned at the face of looking the unit.Use is positioned at a plurality of viewpoints on this face of looking the unit, will look the observability of looking the unit from this and calculate as the combination observability of looking from these points.Distance between the viewpoint is determined to use the size of shrinking with the shelter of guaranteeing the conservative property result.Because the observability that positions on the unit are looked a plurality of in the method sampling, so it does not suppose a priori that fully as seen all elements that are not blocked look from the whole unit of looking.

Compare with many previously described methods (comprising volume observability and generalized projection), the people's such as Wonka method does not suppose that fully as seen all elements that are not blocked look from looking everywhere on the elemental area.Because its sampling observability of looking from a plurality of positions of looking on the unit, so it can be similar to the back projection that never blocks the partial occlusion that element looks that the unit is looked in explanation.Because can cause umbra border, plane (NonE-EV) and than more accurate secondary umbra border (luminophor EEE and non-luminophor EEE), E-EV border depending on the element in the penumbra in light unit/source, the E-EV border is never blocked element and is looked and as seen generate by supposing the whole unit of looking.It is for looking the 2.5D model proposition that the unit is rectangle that the another kind of the method is implemented.Implement to compare the quantity that this has greatly reduced the complexity of shelter shrink process and has reduced fully required viewpoint sample with full 3D.Regrettably, because this enforcement is limited to the 2.5D model, so he can not use in most roaming is used.

The people such as Chhugani (2005) (Chhugani, Jatin, et al. " vLOD:High-Fidelity Walkthrough of Large Virtual Environments. " IEEE Transactions on Visualization and Computer Graphics11.1 (2005): 35-47, its full content merges among the application by reference) another kind of observability pre-computation methods has been described, it utilizes " shelter of contraction " to use the observability of looking from a single point of looking in the unit to be similar to the observability of looking from looking the unit.The method is utilized compound object-space and image-space-wise.Structure " with looking unit and object tangent supporting plane " in object-space.Selection is included in the viewpoint in these supporting planes, and for each supporting plane, structure passes viewpoint and the offset planes parallel with original plane.Said according to the author, the cross-shaped portion of the positive semispace of these offset planes comprises the truncated cone of the actual umbra that is positioned at primary object.For each the object polygon that generates supporting plane, polygonal contraction is to be determined by the corresponding polygonal skew for selected viewpoint.Blocking of shelter object back is then to draw with the occlusion query expansion of depth buffer and hide the thing that is blocked and determine by playing up the contraction version of looking from viewpoint.This inquiry returns zero for the sightless thing that is blocked.The method is carried out limited shelter by the shelter of playing up the thing contraction before that is blocked and is merged.Must generate and play up with identical viewpoint the shelter of all contractions.This viewpoint must be arranged in the truncate cone of all shelters.The volume of the truncate centrum that shrink with maximization the position of selecting viewpoint with protruding two suboptimizations that realize local minimum programme and.

The precision of the shelter that shrinks is to a great extent by the size of the shelter of considering with distribute to determine.Thereby, be not easy in the method control accuracy.

Be connected the non-protruding shelter of polygonal mesh with connection although the method admits to comprise each polygon, it is not suitable for having the shelter in hole.This is that to have actual be each thing that is blocked of the single polyline " border " of the profile contour type of looking from the district because the method depends on.Because general some the large polygonal meshs that block (for example, building) of significantly looking from the district that produce also have a plurality of topological holes (for example, door and window), so this is a significantly restriction.

Analysis according to the front is apparent that, many existing PVS pre-computation methods based on simple E-EV event face (are for example used, generalized projection, volume observability) the conservative linearization to the umbra border approximate, its element of not being blocked of hypothesis is from looking any place on the unit look as seen (that is, as seen the whole element that never blocks the unit of looking looks).

Although the luv space method of existing observability precomputation of looking from the district is not used the visible event boundaries of accurate secondary; The visible event face of secondary in other observability computation luv spaces.These one of are used, observability skeleton (Visibility Skeleton) Durand(1997) be for the data structure of answering the global visibility inquiry.Other application, discontinuous network is formatted, and is the method for calculating the brightness in the area source situation.At first will check the uncontinuity gridding method.

As previously described, at Drettakis and Fuime(1994) completely uncontinuity gridding method in, to appearing at that all visibility events faces between polygon light source and the polyhedral model are identified and the polygon of itself and model being intersected.These intersect " the completely uncontinuity grid " about the source that comprises model.The uncontinuity grid is divided into surface grids with the model geometric image, so that the view in source is equal in topology in each face.The uncontinuity grid is useful data structure for calculating for the overall brightness on umbra border and penumbra border completely.

In uncontinuity gridding method completely, identify four class event faces (seeing Table Ia and Table I b).Two classes in these event faces are that the two classes plane and other are quadric surfaces.Previous two plane event Noodles types discussing, E-EV and NonE-EV use conservatively to comprise the umbra boundary surface of looking from the district by the conservative visibility events method of looking from the district.In some cases, these planes are actually by silhouette edge with as the part of looking the accurate umbra border that the unit forms of light source.

Two types quadric surface: luminophor-Bian-Bian-Bian (luminophor-EEE or EeEE) and non-luminophor-Bian-Bian-Bian (non-smooth body EEE) are the parts of some the visibility events face between area source and the model gengon.For example, in some cases, these quadric surfaces can be by silhouette edge with as the part of looking the accurate umbra border that the unit forms of light source.In most of the cases, these event faces part that is penumbras.The uncontinuity grid method has been described the technology of identifying for to all secondary event faces that appear between area source and the model polygon.

For example, at Drettakis and Fuime(1994) in the discussion done, can come luminophor EEE and non-smooth body EEE event face are identified by between generator limit and the polygonal convex closure of luminophor, forming axle.Luminophor EEE event face is that limit and other limits by the former initial line in this axle, luminophor forms.Non-luminophor EEE event face is to forming by the former initial line in the axle and non-parallel side.Non-luminophor EEE face is the face that those and luminophor polygon intersect.In both cases, the secondary event face that shows recognition rule with the parameter equation on the first generator limit:

P _t=a ₁+t(b ₁-a ₁)

A wherein ₁And b ₁The end points e that is ₁

Comprise a P and e by formation ₂And some P and e ₃Two planes find for the some P on the quadric surface of rule _tThe value of t.Crossing formation and the e on these two planes ₁The line that intersects.

By calculating for limit e ₂End points a ₂And b ₂And for limit e ₃End points a ₃And b ₃T find quadric significant interval of the rule on the generator limit.The cross-shaped portion at interval is the effective district on the first generator limit.(Teller advises that also quadric Parametric Representation of service regeulations represents the face among the 3D.But in the method for Teller, in fact in luv space, there is not the secondary visibility events face of recognition rule.Substitute their delimiter, identification extreme value ruler in the 5D space of lines).

In the uncontinuity gridding method, in case identified quadric surface by the significant interval of seeking its generator limit, then determine the coefficient of corresponding quadratic equation:

Ax ²+By ²+Cz ²+Dyz+Exz+Fxy+Gx+Hy+Iz+J=0

The polygonal cross-shaped portion of this quadric surface and model is quafric curve.Come it is determined so that polygon is embedded in the plane of z=0 by changing three generation limits.The coefficient of the quadratic equation by correspondence deducts all that comprise z and limits quafric curve.In order to generate the uncontinuity grid elements, with this quafric curve and intersect on the polygonal limit of model and use line to scan the observability Processing Algorithm for observability this quafric curve is checked.

In the uncontinuity gridding method, identify comprising polygonal all the visibility events faces of area source and model.These visibility events faces not only comprise umbra and extreme value penumbra border but also are included in the gamut from the topological view of the light source of model geometric shape or many other event faces that " aspect " changes.In the uncontinuity gridding, the visibility events face is identified and intersects with the model polygon, but does not calculate concrete encirclement volume such as the umbra volume of these faces.These intersect general the generation and are difficult to the 5th degree space curve that robust ground solves.Fortunately, use the brightness calculation of uncontinuity grid not need to express the umbra volume.

The structure of uncontinuity grid does not need event face and model polygon to intersect completely, forms line or quafric curve at polygonal.Undertaken these and intersect by passing the model projection surface.Space subdivided data structure with rule-based grid limits the quantity that intersects of carrying out.Calculating after all these intersect, the observability step is determined the visible sub-segmentation on the wedge.Therefore, the structure of uncontinuity grid is not that output is responsive.And the cost that E-EV processes is the O(n of polygon quantity ²).Also process quadric surface by all quafric curves of at first seeking quadric surface and the polygonal crossing formation of model, observability on the quadric surface is to solve by the line swept algorithm of using subsequently, and the cost that quadric surface is processed is the O(n of polygonal quantity ⁶).

With the uncontinuity grid is identical completely, observability skeleton (people 1997 such as Durand) (Durand, Fredo; Drettakis, George; Puech, Calude; " The Visibility Skeleton:a Powerful and Efficient Multi-Purpose Global Visibilty Tool; SIGGRAPH1997Proceedings, its full content merges among the application by reference) be the data structure of explaining secondary visibility events border with the luv space method.This observability skeleton is the full directory of the visible event between the limit that appears in the polyhedron environment.In the observability skeleton, the visibility information of model is organized as following graphic structure: wherein the extreme value ruler is that node and the visibility events face of figure are the arcs of figure.The observability skeleton can be used for answering the observability inquiry in those scenes that for example occur during overall brightness calculation.

Be different from completely uncontinuity gridding, the observability skeleton has been avoided the direct processing to the linearity band that comprises the visible event face of secondary.Alternatively, by only directly calculating encirclement event face self and the extreme value ruler corresponding with the node of observability skeleton diagram structure constructed this skeleton.

Be incident on four limits (EEEE node) the extreme value ruler generally speaking, identify above-mentioned node with the set that is formed on the tetrahedron wedge between these four limits.In the method, between two limits on above-mentioned limit, form the broad sense tetrahedron, as shown in the people's such as Durand (1997) Fig. 8 a, Fig. 8 b and Fig. 8 c.The figure that mentions in this section refers to the figure in the paper of the people such as Durand (1997).In Fig. 8 a, show the broad sense tetrahedron that is formed by limit ei and limit ej.In Fig. 8 b, be illustrated as ej segmentation in the broad sense tetrahedron that is formed by ei and ej along the 3rd limit ej.[the part ej in this broad sense tetrahedron can utilize ei and ej to form secondary visibility events (EEE) face].In Fig. 8 c, show the 4th limit el.The 4th limit is similarly by limiting to three other broad sense tetrahedron: ekej, the ekei that may intersect with it and ejei.The segmentation el that is positioned at all these tetrahedron wedges can form three quadric surfaces with other three limits.Only there is a line in fact to intersect with el and other lines.This is extreme value luminophor-EEE line or the node that comprises the observability skeleton on these four limits.It is by the simple binary chop of limited segmentation el is sought.In this binary chop, select the initial estimation of the point of crossing P on the el.Then, will be intersected with ej and e by the plane that this point and ei form, provide two lines that come from the P place.Cross-shaped portion P on the estimated el comes refinement by binary chop until the angle that comes between two lines at P place approaches zero.When this appears at above-mentioned line congruence, therefore intersect ei, ej, ek and el.The extreme value line and the model polygon that use ray cast so to identify intersect to determine that any scene polygon blocks the extreme value line between its generation limit.If line is so blocked, then do not record the extreme value ruler.

Other nodes of observability skeleton such as EVE, VEE and EEV node form plane visibility events face (for example, limit VE), and by making edge closing intersect to seek above-mentioned node with corresponding broad sense tetrahedron.

The extreme value ruler of so identification is stored as clearly the node of observability skeleton.Directly do not calculate the visibility events face (polygon or quadric surface) that is surrounded by these lines, but alternatively, clearly it is stored as the arc among the figure.The part limit of event face is to derive from the node that is connected to corresponding arc.Use subsequently for the observability skeleton of global visibility inquiry for example in the situation that the uncontinuity gridding of area source, may need example as by Teller(1992) quadric form of described parameter generates quadric surface.

Be apparent that according to aforementioned analysis, uncontinuity gridding and observability skeleton technique comprise for identification by the plane visibility events face of area source generation and the luv space technology of secondary visibility events face.The segmentation that both have used broad sense tetrahedron wedge to test to identify quadric surface and support above-mentioned quadric limit tlv triple effectively.Two kinds of methods produce all the visibility events faces between the dependence edge.Neither one is configured and effectively generates only umbra boundary surface of looking from the district of being correlated with in calculating from the observability that the district looks in two kinds of methods.

The another kind of method of the observability that calculating is looked from the district is that this problem is converted to space of lines and uses Plucker coordinates to calculate the umbra boundary surface.

As previously mentioned, accurate planar element and the necessary computing machine of quadric surface element on the non-penumbra border of entrance sequence are calculated in Teller(1992) method exploitation.The method is transformed into the 5D space of lines with problem.

In the situation that there are not different entrances, to compare with the general considerations of the observability of looking from area source (or equally, depending on the unit), the entrance sequence is more effective visibility problem significantly.In addition, in order to identify the quadric surface element on non-penumbra border, Teller must use Plucker coordinates that problem is transformed into space of lines and carry out lineoid in 5D and intersect.The potential robustness problem that this conversion can not occur when having increased the algorithm complex of processing and having introduced in being operated in luv space.

Since calendar year 2001, two groups of researcher Bittner(2001) (J.Bittner and J.P ˇ rikryl.Exact Regional Visibility using Line Space Partitioning.Tech.Rep.TR-186-2-01-06, Institute of Computer Graphics and Algorithms, Vienna University of Technology, March2001.) and Nierenstein(2002) (Nirenstein, S., E.Blake, and J.Gain. " Exact From-Region Visibility Culling. " Proceedings of the13th Eurographics Workshop on Rendering.Proc.of ACM International Conference Proceeding Series, Pisa, Italy, 2002.Vol.28.Aire-la-Ville:Eurographics Association, 2002.191-202. its full content merges among the application by reference) developed the method for accurately looking the unit of calculating for polygon PVS.Calculate equally with the accurate non-penumbra of Teller, these methods need to be transformed into problem Plucker coordinates and depend on the root searching that comprises svd, robust and the combination of the numerical technique of high-dimensional convex hull computation.Different from the method for Teller, these methods do not need model is divided into the BSP tree with the entrance that lists automatically.

Usually, these two kinds of exact methods, Niernstein(2002) and Bittner(2001) be configured to determine between two protruding graphic primitives (that is, polygon), whether exist the observability inquiry of the sight line that is not blocked.One of polygon of testing is the face of looking the unit, the grid polygon that another polygon of testing is modeling environment.Inquiry determines that the polygon in model blocks all sight lines between the polygon of testing individually or in combination.This occlusion query represents space of lines between the polygon by the 5D Euclidean space of being derived by the Pu Lvke space.This mapping needs singular value matrix to decompose.In subsequent step, the method is used the structure solid geometry graphic operation that carries out in 5 dimension spaces.These processing that form the basis of observability inquiry have high assessing the cost.In addition, because the basic organization of the method has used polygon to polygon to inquire about, so the cost on naturally implementing is the O(n of polygon quantity ^2.15) (Nirenstein2002).

By using the test of ordinary acceptance and ordinary refusal to improve the scalability of the method under this worst case.Polygon to the ordinary acceptance of polygonal observability is established with polygon to polygonal ray projection inquiry.If come from that the ray at a test polygon place reaches other test polygons not with database in any polygon between two parties intersect, then can accept prosaically the observability inquiry.Having lower assessing the cost although the observability inquiry of accurate Pu Lvke space is compared in this inquiry, himself is for the relatively test of costliness of ordinary acceptance.Can accelerate with the database of level ground tissue the ordinary refusal of polygonal cluster.If inquiry determines the bounding box of object and be blocked with respect to the viewpoint unit, then all polygons of comprising of besieged box also are blocked.In addition, the method is processed the bounding box self that is blocked as simple " virtual shelter ".(people 2000 such as Koltun) (Koltun, Vladen, Yiorgos Chrysanthou, and Daniel Cohen-Or. " Virtual Occluders:An Efficient Intermediate PVS Representation. " Proceedings of the Eurographics Workshop on Rendering Techniques2000.London:Springer-Verlag, 2000.59-70 its full content merges among the application by reference).As defined by the people such as Koltun (2000), virtual shelter is not the geometric part of master pattern, but still represents one group of plugging-up line.If the bounding box of object is not blocked, then the bounding box of object can be used as for any geometric shelter that is positioned at its back.Because bounding box self is for being more than sufficient for the test of blocking at the object of its back, so any one polygon is thought of as the shelter candidate in the polygon in the bounding box that does not need to be blocked.Use these virtual shelters by the vertical processing in conjunction with the people's such as Nirenstein (2000) scenario objects, significantly with the measurability of the method for some scenes of testing from O(n ^2.15) bring up to O(n ^1.15).However, the method is shown to have large constant computing cost.For the intensive scale Forest Scene that blocks that is comprised of 7,800,000 triangles, pre-service needs 2 days 0 22 hours on two Pentium IV1.7GHz multiprocessors.In contrast to this, use the generalized projection method that is implemented in the people such as Durand on the 200MHz MIPS R10000 uniprocessor with SGI Onyx2 graphic hardware that identical database is carried out pre-service and only need 59 minutes.This accurate method has proposed average 99.12% geometric figure, by contrast, uses conservative generalized projection method to realize 95.5% rejecting.

A high reason that assesses the cost of this exact method is: polygon is processed by blocking that each polygon causes individually to polygonal occlusion query, and does not consider clearly that annexation between the polygon calculates polymerization and block.This exact method only explains that by the 5D structure solid geometry graphics process of processing individually each polygonal costliness connecting polygonal combination blocks.For this accurate method, only come definite polygonal combination that connects to block by deducting individually each 5D polyhedron (polygonal candidate is blocked in expression) from 5D polyhedron (the expression unit is to polygonal sight line).In the situation of the grid that connects, the limit of sharing represents the ordinary situation that shelter merges, but for exact method, the fusion that must calculate clearly the fusion of these shelters and these shelters represents the degenerate case of algorithm, and this is because result's polyhedron accurately intersects along the limit of sharing.In this scene, the people's such as Niernstein (2002) method has been ignored the major issue of identification on those concrete limits of the visibility events face that supporting potentially of polygon model looked from the district fully, and alternatively uses all polygon limits to carry out the observability inquiry.

In rear a kind of enforcement, the people such as Nirenstein (2005) (Nirenstein, S., Haumont, D., Makinen, O., A Low Dimensioinal Framework for Exact Polygon-to-Polygon Occlusion Queries, Eurographics Sysmposum on Rendering2005, its full content merges among the application by reference) potential solve this shortcoming of the method from looking profile border and the polyhedron that stops along these boundary formation 5D of looking the unit by identifying.The definition of the silhouette edge of looking from the district that uses in the method is identical with the above-mentioned definition of using early people's such as Drettakis (1994) completely uncontinuity gridding method in fact.Although use a testing table of this lifting to implement the method has been accelerated the factor 30, the method still has high constant computing cost.

Except calculating the ground costliness, this accurate method is difficult to robust ground and implements.The svd of the method, the root of robust are sought and more high-dimensional structure solid geometry graphics calculations is often degenerated very responsive to digital tolerance and geometric figure.

Another shortcoming of the method that should look from the district accurately is: the general nonrecognition of current enforcement and remove the polygonal part that is blocked that partly is blocked.Face and the polygon between the model polygon-polygonal observability inquiry of looking the unit used in the current enforcement of the method.If this inquiry specifically is configured to identify the district of not blocking between the polygon of being tested and is detected any such district then premature termination.Such enforcement comprises the whole polygon among the PVS, even it only has sub-fraction to look as seen from looking the unit.Therefore, although can be visible polygonal " accurately " set of looking from the district by these PVS that implement to calculate, PVS can over-evaluate significantly from being used for the polygonal greatly unit face area that exposes of looking visible of looking.This can cause the remarkable overdraw of working time.Revise the enforcement of looking from the district accurately and increase fully the complexity that assesses the cost and implement to determine the segmentation meeting that is not blocked, this be because: 1) can lose the benefit of premature termination, and 2) border between the segmentation that is not blocked and the segmentation that is blocked is quadratic form.

Because these space of lines methods are calculated the secondary umbra border between source polygon and the target polygon, so they can provide the accurate scheme for this observability inquiry.By contrast, the conservative approach of observability precomputation clearly (volume observability) or impliedly (projecting method) use more coarse linearization umbra border.But because these conservative methods are operated in luv space, so compare that the root that needs robust is sought and the space of lines method of more high-dimensional structure solid geometry figure, they are suitable for more simply, the more enforcement of robust.

In generalized projection method and volume observability method and accurate from the method that the district looks, look the unit for the parallelepipedon that comprises the space of to open the navigation or air flight and calculate PVS.Compare the general convex polyhedron that is used by BSP/ entrance sequence method and look the unit, use the plane hexahedron to look the unit and have some advantages.Limit parallelepipedon and look the spatial level (for example, k-d tree) that the space segmentation of unit can easily be arranged to be conducive to the hierarchical method determined for PVS.In the method for being used by generalized projection method and volume visual method, determine PVS and this PVS is used for recursively determining that son lower in hierarchical structure looks the working set of the PVS of unit for the unit of looking that is in high-level place in the hierarchical structure.

Another advantage of parallelepipedon unit is: they have and simple cell abutment relation for adjacent cells.This relation is used in the generalized projection enforcement of implementing increment PVS storage scheme, the people such as Durand (2000).In this scheme, the PVS of unit is looked whole in storage for many keys.Look the unit for most other, the storage expression is in abutting connection with the set of the difference of the PVS that looks the unit.This storage scheme has reduced fully for PVS data storage demand.

In generalized projection was implemented, the PVS that calculates encoded to scene-figure-unit observability to the conservative unit of looking, coarse granular level place.For this approximate solution, for 12,166 look the unit (expression comprise 600 ten thousand polygonal city model the street district 1/12) increment PVS storage need the 60MB storage space.Inferring, can be 720MB for the increment PVS data storage space of looking the unit that comprises all streets.In part working time, all geometric figures are stored in the primary memory and from CD, obtain increment PVS data.

The observability method of having used the another kind of increment PVS storage scheme to look from the district is the people's such as Chhugani (2005) Vold method.In this was implemented, the observability solution of looking from the district used the modification of " contraction shelter " method that the conservative unit-object PVS that looks is provided.

Increment PVS is the tabulation of object ID, and above-mentioned object ID refers to for the recently visible or invisible object recently of looking the unit transition.Compare with the generalized projection method, Vlod implements need to not store all model geometric figures in primary memory.Alternatively, geometric figure is stored on the disk and is used for instructing speculative prefetching to process the viewpoint position of current and prediction, it dynamically loads increment PVS data and model geometric graph data.With object reorder scheme with the model geometric graphics memory on disk, this object object of scheme by will often look ahead together that reorder leaves on the disk together to reduce magnetic disc access times.Also increment PVS data are stored on the disk.For having 1,300,000 triangles and 500,000 propulsion system models of looking the unit, need the storage space of 7GB to store increment PVS object ID.

In working time, Vold implements can be to excessive and model that can not be stored on the primary memory carries out real-time rendering.Because the model of playing up in Vlod implements does not have veining, so the method does not solve storage and the dynamic prefetch of texture information.In the roaming in most of modern times is used as game, for the amount of the texture information of model usually much larger than the amount of geometry information.

The Vlod system is the example of outer (out-of-core) real-time rendering system of nuclear, and this system uses based on the geometric figure of the observability of looking from the district of institute's precomputation and looks ahead.Example (the Database Management for Interactive Display of Large Architectural Models early of the method for Funkhouser, Proceedings of the conference on Graphics interface'96Toronto, Ontario, Canada Pages:1-8Year of Publication:1996ISBN:0-9695338-5-3, its full content merges among the application by reference) used the geometric figure PVS data of calculating by the entrance sequence method.Untextured geometric figure is also used in this enforcement, and the same with Vold, does not solve looking ahead of texture information.

Other external memory methods are used look ahead (for example, preferential layer projection algorithm or PLP) based on the conservative observability method of looking from some working time, and it is used for determining from the look conservative subset (IWALK, MMR) of visible model of viewpoint.In a kind of modification of the method, the processing of primitive re-projection be used for directly identification since viewpoint move become the model geometric figure that recently exposes (U.S. Patent No. 6111582, Jenkins).These observability methods of looking from the district must be calculated in working time, had therefore increased overall operation Time Calculation cost.

The target of the outer rendering system of nuclear is to make it possible to very large, the detailed environment that can not be suitable for core memory is carried out unbroken prospecting.After effectively implementing, this method for stream processing can be eliminated by the caused regular interruption of traditional loading scheme, in traditional loading scheme, and the whole part of loading environment (for example, the outpost of the tax office) before next outpost of the tax office reaches.The 3D model of load is subdivided into loading and the demonstration that different " level " simplified graphical information up hill and dale, forces simultaneously the user to go to experience a series of positions of breaking, these positions were separated by the load time of the continuity that often interrupts experience.

Data available transmission between supplementary storage and the nuclear is to process the significant limitation factor (the Brad Bulkley that implements for stream, " The Edge of the World " Game Developer Magazine June/July2006pg.19, its full content merges among the application by reference).Increment PVS storage scheme can reduce to flow the required transfer rate of processing prefetch data fully.Current increment PVS implements not provide the method for managing grain information.In addition, they use the unit of coarseness to arrive scene figure-unit PVS data to object or unit, calculate above-mentioned data with the coarse observability precomputation of looking from the district that causes over-evaluating PVS/ increment PVS data.If increment PVS data cause prefetch process to surpass transfer rate between supplementary storage and the internal memory, then can produce the observability error.

Compare existing method, can determine that the observability pre-computation methods of looking from the district of the polygon segmentation that is blocked and texture can produce more accurate unit to polygon PVS/ increment PVS.This can reduce to support to flow processes the required transfer rate and can be by reducing to overdraw to strengthen the performance of viewing hardware of looking ahead.

The conclusion of background technology part

Be apparent that according to aforementioned analysis to prior art, existing observability pre-computation methods of looking from the district is used: a) produce the out of true visibility events border that out of true PVS separates or b) must tie up the accurate visibility events face that calculate in the space of lines five.Such space of lines calculates and causes high assessing the cost and algorithm complex and be difficult to the enforcement of robust ground.In addition, for the single collection of polytopic plants, some visibility events faces of looking from the district accurately are to be similar to well by better simply linearization extreme value umbra border; Other then are not.This is so that accurate method is exceedingly responsive to detailed input from the following aspect: in some districts of typical polyhedral model, can spend a large amount of calculating and calculate seldom blocking of amount.

Therefore, the general PVS that Conservative Linear umbra event face in the luv space is identified determines that method estimates these faces departing from from accurate event face, and these faces of adaptively refinement are with the surface of approximate exact more accurately, compare with existing method, said method can carry out the observability precomputation of looking from the district originally with the degree of accuracy that improved and being calculated to be of having reduced.

The PVS of the degree of accuracy of this reality control determines that method can and reduce carrying cost and is conducive to use based on the intermediate representation that the stream of observability is looked ahead in conjunction with increment PVS.Because initially only transmit visible geometric figure, texture and other graphic elements near user's initial position, thus should be based on stream forecasting method meeting of observability so that the user can begin soon with large veined 3D model alternately.This primary data is normally for the sub-fraction of the whole graphic data base of modeling environment.With such as the MPEG-4 part 11(VRML or the X3D that do not specify effectively based on the method for stream processing of looking ahead of observability) etc. existing method compare, the method can reduce significantly for the mutual stand-by period.Existing method like this requires to load whole database usually before beginning alternately, or alternately, suffers observability error (for example, the unexpected appearance of object) during user's navigation.

Summary of the invention

In the exemplary embodiment, method has determined that the grid polygon forms polygonal mesh from the look set of visible grid polygon or the polygonal segmentation of grid of the vision area with a plurality of vision areas summit.The method comprises at least one support polygon of determining between vision area and the polygonal mesh.The method further comprises according to described at least one support polygon constructs at least one wedge, thereby described at least one wedge extends away from vision area and exceeds at least polygonal mesh.In addition, described method comprises one or more cross-shaped portions of determining grid polygon and wedge.Additionally, the method comprises with the polygon one or more cross-shaped portions of determined at least one wedge and grid and determining from the look set of visible grid polygon or the polygonal segmentation of grid of vision area.

In the exemplary embodiment, the method for proposition be used for to receive from the look set of visible grid polygon or the polygonal segmentation of described grid of the vision area with a plurality of vision areas summit, and described grid polygon forms polygonal mesh.The method comprises and receiving from the look set of visible described grid polygon or the polygonal segmentation of described grid of described vision area, the polygonal described set of grid is determined according at least one wedge and the polygonal one or more cross-shaped portions of described grid, described at least one wedge is incident on the single order silhouette edge, and described at least one wedge is to construct according at least one support polygon between described vision area and the described polygonal mesh.The method further comprises showing from the look set of visible grid polygon or the polygonal segmentation of described grid of described vision area described reception.

In the exemplary embodiment, system determines that described grid polygon forms polygonal mesh from the look set of visible grid polygon or the polygonal segmentation of described grid of the vision area with a plurality of vision areas summit.This system comprises having the processor that is configured to determine at least one support polygon between described vision area and the described polygonal mesh.This processor also is configured to construct at least one wedge according to described at least one support polygon, thereby described at least one wedge extends away from described vision area and exceeds described at least polygonal mesh.This processor also is configured to determine described wedge and the polygonal one or more cross-shaped portions of described grid.This processor also is configured to determine from the look described set of visible described grid polygon or the polygonal segmentation of described grid of described vision area with one or more cross-shaped portions of described definite described polygonal mesh and described at least one wedge.This system also comprises the client device that shows for the described set to look from described vision area visible institute grid polygon or the polygonal segmentation of described grid.

In the exemplary embodiment, non-transient state computer-readable recording medium has executable instruction stored thereon, when carrying out described executable instruction by processor, described executable instruction so that processor carry out to determine from the look method of set of visible grid polygon or the polygonal segmentation of described grid of the vision area with a plurality of vision areas summit, wherein, the grid polygon forms polygonal mesh.The method comprises at least one support polygon of determining between vision area and the polygonal mesh.The method further comprises according at least one support polygon constructs at least one wedge, thereby at least one wedge extends away from vision area and exceeds at least polygonal mesh.In addition, the method comprises definite wedge and the polygonal one or more cross-shaped portions of grid.Additionally, the method comprises with one or more cross-shaped portions of determined polygonal mesh and at least one wedge and determining from the look set of visible grid polygon or the polygonal segmentation of grid of vision area.

In the exemplary embodiment, method determines that the grid polygon forms polygonal mesh from the look set of visible grid polygon or the polygonal segmentation of grid of vision area.The method comprises that at least one that determine to be incident on the grid polygon block the border from what vision area was looked, and border has the side of not blocking and the side that is blocked from blocking of looking of vision area at least one.The method further comprises to be determined to add to from the look polygonal quantity of grid of set of visible grid polygon or the polygonal segmentation of grid of vision area, by in the intersection of blocking between border and the polygonal mesh of not blocking that at least one is looked from vision area on the side face of the polygonal mesh that intersects being repartitioned to add the grid polygon.In addition, the method comprises and determines to be blocked the polygonal quantity of the grid that blocks on the border by at least one from what vision area was looked.In addition, the method comprises: when having determined that the polygonal quantity of the grid that adds surpasses the polygonal quantity predetermined threshold of grid be blocked, remove at least one and block the border from what vision area was looked.

In the exemplary embodiment, method receives from the look set of visible grid polygon or the polygonal segmentation of described grid of the vision area with a plurality of vision areas summit, and described grid polygon forms polygonal mesh.The method comprises and receiving from the look set of visible described grid polygon or the polygonal segmentation of described grid of described vision area.The polygonal set of grid does not comprise that at least one that remove from the described set of look from described vision area visible grid polygon or the polygonal segmentation of described grid block the border from what vision area was looked, described at least one the border has the side of not blocking and the side that is blocked from blocking of looking of vision area.When the polygonal quantity of grid of the described set of having determined to add to grid polygon or the polygonal segmentation of described grid surpass by described at least one look from the district block the polygonal quantity predetermined threshold of the grid that blocks on the border time, at least one is blocked the border removes from the described set of look from described vision area visible grid polygon or the polygonal segmentation of described grid.Add an above-mentioned quantity grid polygon by polygonal mesh on the described side that is not blocked and described at least one face of intersection between the border repartitioning described crossing polygonal mesh that blocks of looking from vision area.The method further comprises the described set to look from described vision area visible grid polygon or the polygonal segmentation of described grid.

In the exemplary embodiment, system determines that described grid polygon forms polygonal mesh from the look set of visible grid polygon or the polygonal segmentation of described grid of vision area.This system comprises the demoder with at least one processor that blocks the border of looking from vision area that is configured to determine on described grid polygon incident, described at least one the border has the side of not blocking and the side that is blocked from blocking of looking of vision area.This processing also is configured to determine add from the look polygonal quantity of grid of described set of visible grid polygon or the polygonal segmentation of described grid of described vision area, adds described grid polygon by polygonal mesh on the described side that is not blocked and described at least one face of intersection between the border repartitioning described crossing polygonal mesh that blocks of looking from vision area.This processor also is configured to determine by described at least one polygonal quantity of grid of blocking from the border that vision area is looked.This processor also is configured to: when the polygonal described quantity of the grid that adds surpasses the polygonal described quantity predetermined threshold of grid be blocked, remove described at least one block the border from what vision area was looked.This system also comprises client device, and described client device shows the described set of look from described vision area visible grid polygon or the polygonal segmentation of described grid.

In the exemplary embodiment, non-transient state computer-readable medium has executable instruction stored thereon, when carrying out described executable instruction by processor, described executable instruction is so that processor carry out to determine that described grid polygon forms polygonal mesh from the look method of set of visible grid polygon or the polygonal segmentation of described grid of vision area.The method comprises that at least one that determine to be incident on the grid polygon block the border from what vision area was looked, described at least one the border has the side of not blocking and the side that is blocked from blocking of looking of vision area.The method comprises that further the polygonal quantity of grid that add is looked the set of visible grid polygon or the polygonal segmentation of grid from vision area determines, by repartitioning the face that intersects polygonal mesh and add the grid polygon from the intersection between the border of blocking that vision area is looked not blocking grid polygon on the side and at least one.In addition, the method comprise determine by described at least one block the polygonal quantity of the grid that blocks on the border from what vision area was looked.In addition, the method comprises: when the polygonal quantity of the grid that adds surpasses the polygonal quantity predetermined threshold of grid be blocked, remove described at least one block the border from what vision area was looked.

In the exemplary embodiment, method determines that the grid polygon forms polygonal mesh from the look set of visible grid polygon or the polygonal segmentation of grid of the vision area with a plurality of summits.The method comprises traversal from least one the grid polygon in the polygonal set of described grid and stop described traversal when running at least one single order silhouette edge.The method further comprises at least one wedge on the described single order silhouette edge of structure.In addition, the method comprise determine described at least one wedge with from polygonal at least one cross-shaped portion of another grid in the polygonal set of grid.The method further comprises determines that whether described at least one cross-shaped portion is to block the border from what vision area was looked.The method also comprise when determined described at least one cross-shaped portion be from vision area look block the border time, continue traversal in the side that is not blocked of blocking the border of looking from vision area.

In the exemplary embodiment, method receives from the look set of visible grid polygon or the polygonal segmentation of described grid of the vision area with a plurality of summits, and described grid polygon forms polygonal mesh.The method is included in the client device place and receives from the look described set of visible grid polygon or the polygonal segmentation of described grid of described vision area.Determined the set that receives by scrambler according to the traversal method that includes following processing: traversal is from least one the grid polygon in the polygonal described set of grid, when running at least one single order silhouette edge, end traversal, at described at least one wedge of single order silhouette edge structure, determine described at least one wedge with from polygonal at least one cross-shaped portion of another grid in the polygonal described set of grid, determine that whether described at least one cross-shaped portion is to block the border from what vision area was looked, and when described at least one cross-shaped portion be from vision area look block the border time, continue described traversal in the described side of not blocking of blocking the border of looking from vision area.The method further comprises by the described set of described client device to look from described vision area visible grid polygon or the polygonal segmentation of described grid and showing.

In the exemplary embodiment, system determines that described grid polygon forms polygonal mesh from the look set of visible grid polygon or the polygonal segmentation of described grid of the vision area with a plurality of summits.This system comprises having the polygonal processor of at least one grid that is configured to travel through from the polygonal described set of grid.This processor also is configured to end described traversal when running at least one single order silhouette edge.This processor also is configured at described at least one wedge of the first silhouette edge structure.This processor also be configured to determine described at least one wedge with from polygonal at least one cross-shaped portion of another grid in the polygonal described set of grid.This processor is configured to also to determine that whether described at least one cross-shaped portion is to block the border from what vision area was looked.This processing also be configured to when determined described at least one cross-shaped portion for look from vision area block the border time continue described traversal in the described side that is not blocked of blocking the border of looking from vision area.This system also comprises for showing from the look client device of set of visible grid polygon or the polygonal segmentation of described grid of vision area.

In the exemplary embodiment, non-transient state computer-readable recording medium has instruction stored thereon, when carrying out described instruction by processor, described instruction is so that this processor determines that described grid polygon forms polygonal mesh from the look method of set of visible grid polygon or the polygonal segmentation of described grid of the vision area with a plurality of summits.The method comprises traversal from least one the grid polygon in the polygonal set of grid and end described traversal when running at least one single order silhouette edge.The method further is included at least one wedge of structure on the described single order silhouette edge.In addition, the method comprise determine described at least one wedge with from polygonal at least one cross-shaped portion of another grid in the polygonal described set of grid.The method further comprises determines that whether described at least one cross-shaped portion is to block the border from what vision area was looked.The method further comprise when determined described at least one cross-shaped portion be from vision area look block the border time continue described traversal in the described side that is not blocked of blocking the border of looking from vision area.

In the exemplary embodiment, method is determined and first group of polygonal information of grid is described in storage, and described grid polygon forms polygonal mesh, and as seen first group of grid polygon look from the first vision area.The method comprise determine and storage from the second vision area second group of grid polygon of looking visible, the second vision area comprises the first vision area.The method further comprises to be determined and the polygonal information of storage at least one grid of expression, and described at least one grid polygon belongs to the polygonal set of grid of connection, and as seen the polygonal described set of the grid of connection looks from the first vision area.The method also comprises the information of at least one boundary edge of polygonal at least one set of grid of determining and storing that expression connects.

In the exemplary embodiment, method is determined first group of grid polygon with institute's canned data, and described grid polygon forms polygonal mesh, and first group of grid polygon comes since the first vision area described grid polygon of looking visible.The method comprises retrieval from the second vision area second group of grid polygon of looking visible, and described the second vision area comprises described the first vision area.The method further comprises the polygonal information of retrieval at least one the first grid of expression, and described at least the first grid polygon belongs to the polygonal set of grid of connection, and as seen the polygonal described set of the grid of connection looks from described the first vision area.The method further comprises at least one boundary edge of polygonal described at least one set of grid that the retrieval expression connects.The method comprises that also traversal belongs to polygonal described at least one set of grid of described second group of polygonal connection of grid.Described traversal comprises: begin described traversal and end described traversal when running into described at least one boundary edge during described traversal at described at least one first grid polygon place.

In the exemplary embodiment, system determines and first group of polygonal information of grid is described in storage, and described grid polygon forms polygonal mesh, and as seen described first group of grid polygon look from the first vision area.This system comprises having and is configured to determine that described the second vision area comprises described the first vision area from the look demoder of visible second group of polygonal processor of grid of the second vision area.This processor also is configured to determine the polygonal information of at least one grid of expression, and described at least one grid polygon belongs to the polygonal set of grid of connection, and as seen the polygonal described set of the grid of connection looks from described the first vision area.This processor also is configured to determine to represent the information of at least one boundary edge in polygonal at least one set of the grid that connects.This system also comprises the described second group of grid polygon of storage, the expression polygonal described information of described at least one grid and represents the database of the described information of described at least one boundary edge.

In the exemplary embodiment, non-transient state computer-readable recording medium has executable instruction stored thereon, when carrying out described executable instruction by processor, described executable instruction is determined and first group of polygonal method of grid of storage description described grid polygon formation polygonal mesh so that described processor is carried out.The method comprise determine and storage from the second vision area second group of grid polygon of looking visible, the second vision area comprises the first vision area.The method further comprises to be determined and the polygonal information of storage at least one grid of expression, and described at least one grid polygon belongs to the polygonal set of grid of connection, and as seen the polygonal set of the grid of connection looks from the first vision area.The method also comprises the information of at least one boundary edge of polygonal at least one set of grid of determining and storing that expression connects.

In the exemplary embodiment, carry out the method for looking ahead and sending for the navigation that control sends to client device with graphical information at server, described graphical information comprises looks visible and from the first vision area the first block graphics information of looking sightless from the second vision area, and described graphical information comprises looks visible and from the first vision area the second block graphics information of looking sightless, the level of detail of the second block graphics information is lower than the level of detail of the first block graphics information from the second vision area.The method comprises to be determined the very first time section that is scheduled at client device to access during described the first block graphics information arrives after the block graphics information on the described client device.The method sends the second block graphics information during further being included in very first time section.

In the exemplary embodiment, carry out control chart shape information from the method that receives based on looking ahead of navigation of server apparatus at client device, described graphical information comprises looks visible and from the first vision area the first block graphics information of looking sightless from the second vision area.The method comprises to be determined the very first time section that is scheduled to access at described client device during described the first block graphics information will arrive after described the first block graphics information on the described client device.The method is accepted adjustment information during further being included in described very first time section, and described adjustment information is for reducing viewpoint speed.

In the exemplary embodiment, the transmission of looking ahead based on navigation of system's control graphical information, described graphical information comprises looks visible and from the first vision area the first block graphics information of looking sightless from the second vision area, and described graphical information comprises looks visible and from described the first vision area the second block graphics information of looking sightless, the level of detail of described the second block graphics information is lower than the level of detail of described the first block graphics information from described the second vision area.This system comprises having the server that is configured to following processor: the very first time section that is scheduled to access at described client device during described the first block graphics information will reach after described the first block graphics information on the described client device is determined.This processor also is configured to send described the second block graphics information during described very first time section.This system further comprises be used to the client device that shows described the second block graphics.

In illustrative embodiments, non-transient state computer-readable recording medium has executable instruction stored thereon, when carrying out described executable instruction by the processor in the server, described executable instruction is so that this processor carries out the method for looking ahead and sending based on navigation be used to the graphical information that controls to client device, described graphical information comprises looks visible and from the first vision area the first block graphics information of looking sightless from the second vision area, and described second graph information comprises looks visible and from described the first vision area the second block graphics information of looking sightless, the level of detail of described the second block graphics information is lower than the level of detail of described the first block graphics information from described the second vision area.The method comprise to be scheduled at described client device with the first block graphics information after the first block graphics information on the access client can arrive during very first time section determine.The method sends described the second block graphics information during further being included in described very first time section.

Thereby at least a embodiment has following target and advantage.Below purpose and the tabulation of advantage not exhaustive but be used for emphasizing a kind of only some purposes and advantage of embodiment.

A purpose provides a kind of method of definite the following: a) look potentially visible graphic primitive collection and b each unit from the set of looking the unit) each unit from look unit set potentially part of the visible primitive that partly is blocked of looking.

Advantageously, the unit that the existing method by the observability precomputation of comparing is determined to the unit or the unit to the PVS of primitive, by determining section the part that is blocked of visible primitive, the potential visible collection (PVS) that causes is more accurate.

Another purpose provides a kind of method that does not need represented database is carried out definite PVS of BSP auto Segmentation.

Advantage of the present invention is that it is unpractical complex model that the present invention can be applied to automatic division of BSP.

Another purpose provides the method that does not need problem is transformed into definite PVS of space of lines.

Advantageously, by the solution in the luv space, below having avoided, this method assesses the cost and robustness problem: more high-dimensional constructing stereo geometric operation and quadratic programming that the space of lines method of observability precomputation is required.

Another purpose provides the method for determining PVS from looking the observability precomputation technology of looking the unit of using, described observability precomputation utilization conservative linearization umbra event face, its for accurate event face or conservative property be similar to accurate event face (it can be quadric surface).

Advantageously, can determine with simple test the deviation of concrete linearization umbra event face and corresponding accurate umbra event face.Further advantageously, this method comprises that the single order backprojection that can be used for the conservative linearization umbra event face of adaptively refinement can be the accurate event face of secondary so that conservative linearization umbra event face is focused at.Also further advantageously, in the situation that do not cause that calculating clearly quadric computational complexity can realize this precision.

Another purpose provides a kind of employing the secondary Bian-Bian-polygonal approximation of Bian visibility events face is determined the method for PVS.

Advantageously, compare the PVS that produces by the observability pre-computation methods based on shade of only using summit, plane-Bian visibility events face, the PVS that the embodiment mode produces is more accurate.

Another purpose provides a kind of method of identifying definite PVS of conservative property linearization umbra event face, these faces and the deviation of accurate event face are estimated, and conservative linearization umbra event face is carried out refinement with the event face of approximate exact more accurately.

Advantageously, by only just conservative solution being carried out refinement when approximate conservative solution departs from accurate the solution significantly, the precision of the method can be suitable with the precision of existing observability method of looking from the district accurately, and cost is lower.

Another purpose provides the method for following a kind of definite PVS: by the maximum difference of the umbra volume that produced by secondary event face and polygon event face is estimated and the refinement polygonal approximation to realize the precision of expectation, control the precision to the polygonal approximation of secondary visibility events face.

Advantageously, the complexity of quadric polygonal approximation, directly reflected quadric non-linear or " crooked " that is similar to.

It is effective vertical polygonal mesh ergodic algorithm that another purpose provides observability computation organization, the geometric processing that it is blocked restriction.

Advantage of the present invention is: by implementing vertical trellis traversal, block the border and stop traversal blocking boundary when running into when blocking the border structure, the method has realized the output selectivity properties of blocking environment for intensive.

Another purpose provides following PVS and determines method: do not rely on the existence of protruding shelter but directly utilize intrinsic continuity, the not shape of considering gridding of blocking in the connectedness of stream shape (manifold) grid.

Advantage is: compare such as volume observability, extended projection and depend on the Weiler-Atherton algorithm etc. of uncommon large protruding shelter in real model, the method is more effective aspect blocking in identification.

Another purpose is by the difference among the PVS of adjacent cells being encoded represent the PVS information for each observability unit.

Therefore advantage is: poor (the increment PVS) that the PVS information table be shown the PVS of adjacent unit utilized the visible continuity of place of 3D scene, reduced significantly for the memory requirement of PVS information and supports the stream required message data rate of looking ahead.

Another purpose is that the increment PVS in the middle of utilizing represents, it by which (profile of looking from looking the unit) limit of storage expression comes the difference of looking the PVS between the unit is encoded in the information of looking the umbra border that unit transition period generation causes obviously blocking or expose.

Advantage is: storage reduced increment PVS memory requirement for the silhouette edge that blocks significantly/expose and make it possible to by existing (from adjacency or looking of comprising the unit looks) traversal constructs rapidly in real time to increment PVS data or PVS data.

Another purpose provides the client-server method that following figure shows: the graphic primitive that server stores is all and the surface information that is associated (as, texture information), include model and PVS data and increment PVS data; And client is only stored from the subset of the required model information of the demonstration of the set of looking the unit that can reach from current view point.Client models information is by carrying out dynamic prefetch with recently visible/sightless geometric figure of looking increment PVS corresponding to elementary boundary grouping and texture information and safeguard being encoded to.

Advantage is: the as required flow transferring method based on observability of the picture material during mutual roaming is used is provided, and it does not need the required low delay of the flow transferring method based on compression of images of prior art to connect.

Additional advantage is: the as required flow transferring method based on observability that the picture material during mutual roaming is used is provided, it is mutual that the required bandwidth of the flow transferring method based on compression of images compared to existing technology, the method need significantly less bandwidth to transmit with high resolution.

Another purpose is by the set of safeguarding the geometric figure be presented on the client and surface information at server and by checking during looking ahead that this information to eliminate redundant transmission, reduces increment PVS data are offered the bandwidth of client fully.

Another purpose be by with some levels of geometric figure and face details model information is stored on the server and during the time period of the visible continuity of low space-time (for example, recently the visible two-forty of showing out) the lower level of detail information of transmission and during the time period of the visible continuity of higher space-time the higher level of detail information of transmission.

Advantage is: transmission during the time period of the low visible continuity of space-time relatively on the low level of detail recently visible model information reduced fully to provide increment PVS information required bandwidth period at these other " peak value " bandwidth demands, simultaneously, owing to being lowered and especially being lowered for the face that recently exposes during the time period of low space-time continuity at human vision sensitivity, produced the image stream that can't harm in the perception.

Advantage is: be in the available bandwidth during visible model information makes it possible to effectively utilize these other " running through " bandwidth demand time periods recently on the relatively high level of detail in transmission during the time period of the visible continuity of high space-time, simultaneously, by replacing low level of detail information in the time period of fully differentiating the required time match of the details of adding with the human visual system with higher level of detail information, produce graphical stream harmless in the perception.

Another purpose provides a kind of method of nuclear outer (out-of-core) of figure demonstration: supplementary storage is stored in all graphic primitives and comprises the surface information that is associated (for example, texture information) of model and PVS data; And the nuclear storer is only stored for the subset from the required model information of the demonstration of the set of looking the unit that can arrive from current view point.Nuclear memory model information is by visible/sightless geometric figure recently and the dynamic prefetch that is encoded as the texture information of the increment PVS grouping corresponding with looking elementary boundary are safeguarded.

Advantage is that embodiment provides the conventional method of looking ahead based on observability stream of having got rid of following demand: manually the database with modeling at random is divided into different levels or subhierarchy, these different levels once must load one, based on the event or the location trigger that manually are arranged in by hierarchical Design person in the model above-mentioned subhierarchy stream are processed kernel memory.

Additional advantages of the present invention are: process by the stream of looking ahead based on observability that fine granularity is provided, the method is conducive to design following game and realistic model database: by have a mind to structure observability " bottleneck " its can be more continuously and burden not, and it can not produce and processes the user that coarseness subhierarchy data cause by stream and experience pause.

Another purpose is to be stored in model information in the supplementary storage and in the visible continuity of low space-time (for example by the many levels with a plurality of geometric figures and face details, recently the lower level of detail information of transmission and in the higher level of detail information of phase time period of the higher visible continuity of space-time transmission during the time period visible two-forty of showing out) reduces fully with increment PVS data transmission required bandwidth of nuclear storer in the rendering system outside nuclear.

Advantage is: the recently visual model information that is in relatively low level of detail in transmission during the time period of the low visible continuity of space-time reduced to provide the increment PVS bandwidth of information fully during these other " peak value " bandwidth demand time periods, simultaneously, because the human vision susceptibility generally is reduced and especially especially is reduced for the face that recently exposes during the time period of low space-time continuity, so produced the image stream that can't harm in the perception.

Other advantages instructions according to the present invention is clearly.

Description of drawings

Fig. 1 shows the exemplary process diagram with the top-down tissue of the linearization umbra event face that rotation and the method for scanning are come at single order profile grid silhouette edge place or place, summit structure is conservative or wedge; This process flow diagram shows by constructing corresponding SE-ME wedge and identifies clearly and the degenerate case of managing parallel support and look element sides and silhouette edge;

Fig. 2 A shows the exemplary diagram of looking the unit and having two polygonal meshs that are incident on two single order wedges on the single order silhouette edge;

Fig. 2 B show look the unit and have (back projection) single order wedge of looking from silhouette edge and corresponding high-order from the exemplary diagram of two polygonal meshs of (orthogonal projection) wedge of looking the unit and looking;

Fig. 3 shows look from the district exemplary process diagram of method of silhouette edge of (looking from looking the unit in this case) of identification single order; Fig. 3 shows the details of the step 110 among Fig. 1;

Fig. 4 A shows the exemplary process diagram that structure is incident on the method for the SV-ME support polygon on the grid silhouette edge; Fig. 4 A has provided the additional detail in the processing shown in the step 116 of Fig. 1;

Two candidate's support polygons that Fig. 4 B shows mesh object M1, looks the unit and have their anglecs of rotation separately;

Whether Fig. 4 C shows be used to the polygon of determining to be formed on the single order silhouette edge and look between the summit, unit is the exemplary process diagram of the test of support polygon;

Fig. 4 D1 shows two polygonal exemplary diagram of grid with consistent summit ordering;

Fig. 4 D2 shows two polygonal exemplary diagram of grid with the ordering of inconsistent summit;

Fig. 5 A shows the exemplary process diagram of scanning the method that triangle constructs to being incident on SE-MV on the inside corner grid profile summit;

Fig. 5 B is the continuation of Fig. 5 A;

Whether Fig. 5 C shows be used to the polygon of determining to be formed on inside corner single order profile summit and look between the element sides is the exemplary process diagram of the test of support polygon;

Fig. 6 A shows the exemplary process diagram of constructing the method for SV-ME and SE-ME wedge according to the SV-ME of correspondence and SE-ME support polygon;

Fig. 6 B shows the exemplary process diagram according to the method for the SE-MV support polygon structure SE-MV wedge of correspondence;

Fig. 7 A shows protruding exemplary diagram of looking unit and non-convex polygon grid, and the single order of grid is looked and illustrated with heavy line from looking the unit silhouette edge, and this figure is along from looking the unit towards the skeleton view of the general direction of polygonal mesh;

Fig. 7 B1 shows the exemplary diagram of the object identical with object among Fig. 7 A, but this figure is along the skeleton view from polygonal mesh towards the general direction of looking the unit;

Fig. 7 B2 shows the polygonal mesh different from the polygonal mesh shown in Fig. 7 B1, and what show grid is not the inside corner limit of single order silhouette edge;

Fig. 7 C1 shows the exemplary diagram for the support polygon of single order silhouette edge A and B, and this figure is along from looking the unit towards the skeleton view of the general direction of mesh object;

Fig. 7 C2 shows the exemplary diagram for the support polygon of single order silhouette edge A and B and corresponding source summit-Grid Edge (SV-ME) wedge, and this figure is along from looking the unit towards the skeleton view of the general direction of mesh object;

Fig. 7 C3 is the exemplary diagram that only shows the SV-ME wedge that the extension according to the limit of the support polygon of correspondence forms;

Fig. 7 D1 shows the exemplary diagram of the object identical with the object shown in Fig. 7 C, but this figure is the skeleton view from mesh object towards the general direction of looking the unit;

Fig. 7 D2 shows the exemplary diagram of the object identical with the object shown in Fig. 7 C1, but this figure is the skeleton view from mesh object towards the general direction of looking the unit;

Fig. 7 D3 shows the exemplary diagram of the object identical with the described object of Fig. 7 C2, but this figure is the skeleton view from mesh object towards the general direction of looking the unit;

Fig. 7 D4 shows with the polygonal mesh shown in Fig. 7 D3 and looks the identical polygonal mesh in unit and look the unit and show two hiding figure that rotate wedges that intersect at the place, external corners summit of single order profile contour;

Fig. 7 D5 shows with the polygonal mesh shown in Fig. 7 D4 and looks the unit and polygonal mesh that limited rotation wedge is identical and look the unit and the hiding figure of limited rotation wedge, but is with different visual angles;

Fig. 8 A1 shows the exemplary diagram of scanning triangle (SE-MV support polygon) on the inside corner summit of being shared by the single order silhouette edge that is labeled as A and B; This figure is along from looking the unit towards the skeleton view of the general direction of polygonal mesh object;

Fig. 8 A2 shows the exemplary diagram of the SE-MV wedge that scans triangle (SE-MV support polygon) and correspondence on the inside corner summit of being shared by the single order silhouette edge that is labeled as A and B.This figure is along from looking the unit towards the skeleton view of the general direction of polygonal mesh object;

Fig. 8 A3 shows the exemplary diagram of the inside corner summit of being shared by the single order silhouette edge that is labeled as A and B and corresponding SE-MV wedge; This figure is along from looking the unit to the skeleton view of the general direction of polygonal mesh object;

Fig. 8 A4 shows the exemplary diagram that is incident on the single order wedge that comprises two SV-ME wedges and single SE-MV wedge on silhouette edge A and the B, and all these wedges intersect at the place, inside corner profile summit that is labeled as ICSV.This figure is along from looking the unit to the general direction of polygonal mesh object;

Fig. 8 B1 shows the exemplary diagram of the object identical with the object shown in Fig. 8 A1, but this figure is along the skeleton view from mesh object towards the general direction of looking the unit;

Fig. 8 B2 shows the exemplary diagram of the object identical with the object shown in Fig. 8 A2, but this figure is along the skeleton view from mesh object towards the general direction of looking the unit;

Fig. 8 B3 shows the exemplary diagram of the object identical with the object shown in Fig. 8 A3, but this figure is along the skeleton view from mesh object towards the general direction of looking the unit;

Fig. 8 B4 shows the exemplary diagram that is incident on the single order wedge that comprises two SV-ME wedges and single SE-MV wedge on silhouette edge A and the B, and all these wedges intersect at the place, inside corner profile summit that is labeled as ICSV; This figure is along the skeleton view from polygonal mesh towards the general direction of looking the unit;

Fig. 8 C shows the exemplary diagram on the single order umbra border that is incident on silhouette edge A and the B, and this figure is along from looking the unit to the skeleton view of the general direction of mesh object;

Fig. 9 A shows the Teller(1992 by prior art) exemplary diagram that is incident on the single order umbra border on silhouette edge A and the B of method construct, this figure is along from looking the unit towards the skeleton view of the general direction of mesh object;

Fig. 9 B shows the exemplary diagram of the object identical with the object shown in Fig. 9 A, but this figure is along the skeleton view from mesh object towards the general direction of looking the unit;

Fig. 9 C shows the exemplary diagram on the more accurate umbra border, umbra border that is produced by the Teller method of prior art by comparing of producing of this method, and this figure is along looking the skeleton view of unit towards the general direction of mesh object;

Fig. 9 D shows the exemplary diagram of the object identical with the object shown in Fig. 9 C, but this figure is along the skeleton view from mesh object towards the general direction of looking the unit;

Figure 10 A shows by the exemplary diagram for some additional UBP of the crossing umbra boundary surface that forms of the UBP of some adjacent single order silhouette edges, and this figure is along from looking the unit towards the skeleton view of the general direction of mesh object;

Figure 10 B is with the polygonal mesh shown in Figure 10 A and looks the identical polygonal mesh in unit and look the view of unit, but shows the one group of UBP that forms PAU.

Figure 11 A shows in the situation that the exemplary diagram of the single order visibility events face (wedge) that the compound-contoured contour is generated by this rotation and the method for scanning;

Figure 11 B is the different views of the structure identical with the structure shown in Figure 11 A;

Figure 11 C shows a part that forms continuous linearization umbra event face with at least one high-order rotation wedge at place, compound-contoured summit; Identical with the view of Fig. 2 B and Figure 11 A;

Figure 12 shows with the rotation of wedge and scans the exemplary process diagram of the method for constructing conservative first-order linear umbra uncontinuity grid;

Figure 13 shows and identify and solve the exemplary process diagram of the processing of overlapping circulation during the 3D trellis traversal;

Figure 14 shows the exemplary process diagram of processing about wedge, from the control of the method for looking the element 2D observability figure that looks the unit for constructing with the 2D trellis traversal;

Figure 15 shows for constructing with the 2D trellis traversal about wedge, from the exemplary process diagram of the main traversal processing of the method for looking the element 2D observability figure that looks the unit;

Figure 16 shows for determining that whether 2D uncontinuity net point otherwise conservatively look the look exemplary process diagram of the processing that is blocked of unity element (VCE) from the correspondence of wedge;

Figure 17 shows the exemplary diagram of processing for the control of the method for structure high-order wedge line, and above-mentioned high-order wedge line is used for determining about looking the observability figure of element sides by backprojection;

Figure 18 shows the main exemplary process diagram of processing of the 2D trellis traversal master processing of looking from the summit for the back projection of structure high-order wedge line;

Figure 19 shows with the 3D polygonal mesh and travels through to construct the exemplary process diagram that the control of the output sensitive method of the observability figure that looks from the district is processed;

Figure 20 A shows the main exemplary process diagram of processing of constructing the output sensitive method of the observability figure that conservative linearization looks from the district with the 3D trellis traversal;

Figure 20 B shows the exemplary process diagram of the estimated value of estimating on the inside corner summit of being scanned the difference of the umbra volume that method and intersecting plane method produce by rotation; This difference is used for determining to construct at place, inside corner summit the method for continuous umbra event face;

Figure 20 C to Figure 20 J shows the step of the 3D trellis traversal of polygonal mesh;

Figure 20 K shows the figure around polygonal mesh that comprises other polygonal meshs;

Figure 21 A determines that the segmentation of uncontinuity grid is whether from looking the unit exemplary process diagram of method of (that is, whether the segmentation of uncontinuity grid was looked from the district and blocked the border) of otherwise being blocked of looking;

Figure 21 B is the continuation of Figure 21 A;

Figure 21 C shows the exemplary process diagram of the method that the PVS polygon is categorized as strong visible, unshielding and is just facing all the time;

Figure 22 shows the exemplary process diagram that the control of method of the 3D trellis traversal of the observability figure that the structure back projection looks from silhouette edge is processed, and above-mentioned observability figure is used for determining that the visible support from silhouette edge is looked looks summit, unit (VSVV) and as seen look unit profile contour (VSVSC);

Figure 23 shows the main exemplary process diagram of processing of method of the 3D trellis traversal of the observability figure that the structure back projection looks from silhouette edge, and above-mentioned observability figure is used for determining looking summit, unit (VSVV) and unit profile contour (VSVSC) is looked in visible support from the visible support that silhouette edge is looked;

Figure 24 A shows and determines that dm_segment is the exemplary process diagram of the processing that otherwise is blocked from the silhouette edge source of looking, and above-mentioned silhouette edge source is used for constructing the observability figure back projection of looking from silhouette edge with the 3D trellis traversal;

Figure 24 B is the exemplary continuation of Figure 24 A;

Figure 24 C shows and uses the back projection observability figure that looks from silhouette edge to construct the conservative visible exemplary process diagram that supports the method for looking unit profile contour (VSVSC) that comprises the VSVS corresponding with adjacent silhouette edge;

Figure 25 shows the exemplary process diagram that the point that uses single order wedge and high-order wedge blocks method of testing;

Figure 26 shows and uses the 3D trellis traversal to construct the exemplary process diagram of alternate embodiments method of the method for polyhedron polymerization umbra (PAU) according to umbra boundary polygon (UBP);

Figure 27 A shows the exemplary diagram of looking unit and two polygonal mesh object MESH E and MESH D; It may be coarse in the following segmentation from the SV-ME umbra wedge that the district looks that Figure 27 A shows single order: corresponding support polygon and to look the unit crossing with the geometric figure that supports between the single order silhouette edge;

Figure 27 B shows except the following exemplary diagram of the view identical with the view of Figure 27 A described: the corresponding segments by segmentation single order silhouette edge and use the out of true part of carrying out single order back projection refinement single order wedge as the sub-segmentation of line source; The result is that the out of true part of this wedge is replaced by two SV-ME wedges that connect by single SE-MV wedge, above-mentioned single SE-MV wedge and two SV-ME wedges have formed continuous umbra face together, and it more accurately has been similar to the secondary umbra event face of the reality in the out of true segmentation that is incident on the single order silhouette edge;

Figure 27 C shows except the following exemplary diagram of the view identical with the view of Figure 27 B described: be subdivided into 4 sub-segmentations rather than two segmentations that come the out of true part of the original single order wedge of refinement by the corresponding segments with the single order profile now, and more approximate to produce umbra event face (quadric surface) actual in this district;

Figure 27 D be with different views (from look the unit a little partially after) exemplary diagram of the structure identical with the described structure of Figure 27 A is shown, its show single order profile with segmentation SE1U and SE1O from look the unit look into single order as seen;

Figure 28 shows by checking single order wedge and the accurate maximum possible deviation between the wedge and being the exemplary process diagram that the method for the back projection processing of looking from the limit is controlled in the segmentation of coarse silhouette edge by identification single order wedge;

Figure 29 shows by checking single order wedge and the accurate maximum possible deviation between the wedge and being the exemplary process diagram that the back projection processing of looking from the limit is controlled on coarse simple and compound inside corner profile summit by identification single order SE-MV wedge;

Figure 30 A show identification have among the observability figure of high effective static barriers (ESO) from look the unit look the district that is blocked method and conservatively simplify Occluded boundary and the exemplary process diagram of the processing of corresponding grid profile contour;

Figure 30 B is the continuation of Figure 30 A;

Figure 30 C is the continuation of Figure 30 B;

Figure 30 D shows the 3D that looks unit and two polygonal meshs and hides line chart;

Figure 30 E shows the skeleton view identical with the skeleton view of Figure 30 D and comprises the blocked area and the corresponding hiding line chart of the 3D that blocks the border;

Figure 31 A shows the employed example data structure of mark silhouette edge method;

Figure 31 B is the continuation of Figure 31 A;

Figure 31 C is the continuation of Figure 31 B;

Figure 31 D shows the figure for the data structure of the illustrative embodiments that adopts increment G+ data;

Figure 32 A shows and uses the method on the limit of contour identification contour and summit of exemplary process diagram come to(for) the data structure of mark profile contour;

Figure 32 B is the continuation of Figure 32 A;

Figure 33 A shows identification for from looking unit A to the exemplary process diagram of the method in the poor visibility increment district of the transition of looking unit B;

Figure 33 B is the exemplary continuation of Figure 33 A;

Figure 34 A shows the mark profile contour information of using for single contour and constructs rapidly the exemplary process diagram that observability figure blocks the method for boundary segmentation in working time;

Figure 34 B is the continuation of Figure 34 A;

Figure 35 A shows the exemplary process diagram of the observability figure that derives from the single silhouette edge of mark profile contour being blocked the method that boundary segmentation constructs;

Figure 35 B is the continuation of Figure 35 A;

Figure 36 shows the exemplary process diagram of the processing that the runtime processing of using the ROI border of constructing according to the mark profile contour of pre-stored to construct observability figure ROI is controlled, and wherein the ROI border limits 3D trellis traversal working time simplification, hint of delimiting traversal ROI;

Figure 37 processes the main processing of the seed triangle tabulation of next mark profile contour information structuring ROI according to pre-stored and the assembly of the connection that is connected ROI with 3D trellis traversal working time of simplifying, hinting;

Figure 38 show with the exemplary process diagram of the method on the beginning border that is attached to corresponding mark profile contour for increment G+ sub-grid corresponding to the specific grid elements of recently exposing of looking the unit transition;

Figure 39 A shows exemplary simple shelter;

Figure 39 B show when from connect look unit A and B and observe the time form the exemplary increment district (DR) of blocking by simple (Figure 39 A's) shelter;

Figure 40 shows the observability figure that the identical unification of observability figure of looking from the district with the unification shown in Figure 39 B is looked from the district except the following stated: the part that is positioned at OCCLUSION REGION VIEWCELL B outside of OCCLUSION REGION VIEWCELL A is marked as DR _OThe increment district of blocking of BA(from B to A) and DR _EThe increment district of the exposure of AB(from A to B);

Figure 41 A shows the identification CSV of the observability method (Figure 14, Figure 15 and Figure 16) about wedge and is configured to the exemplary diagram of purposes of the wedge line of SV-ME wedge, and Figure 41 A shows the situation that does not have cuspidated simple CSV;

Figure 41 B shows the identification CSV of the observability method (Figure 14, Figure 15 and Figure 16) about wedge and is configured to the exemplary diagram of purposes of the wedge line of SV-ME wedge; Figure 41 B shows the situation of the degeneration CSV at the tip that forms single order profile contour;

Figure 41 C shows the illustrative plot that is incident on the SE-ME wedge that intersects with 3 polygonal mesh objects on the single order silhouette edge, shows single order from the wedge line (WL) looking element sides and look and they and the polygonal cross-shaped portion of grid; This figure is used for the operation that explanation is used for constructing the 2D trellis traversal processing of the observability figure (Figure 15 and relevant figure) about wedge;

Figure 41 D shows polygonal mesh, looks the skeleton view of the part that comprises tip and compound-contoured summit of unit and single order profile contour;

Figure 42 A shows and uses level to look the exemplary process diagram of the method for unit;

Figure 42 B shows and uses level to look the exemplary process diagram of the method for unit;

Figure 43 A is the exemplary diagram that the data structure of safeguarding for the increment VM/PVS that uses increment VM/PVS data is shown;

Figure 43 B is the continuation of Figure 43 A;

Figure 44 A shows and support to use the increment VM/PVS(increment G+ sub-grid that sends from remote server) exemplary process diagram of the data storage safeguarded of the increment VM/PVS of data and transmission method;

Figure 44 B is the continuation of Figure 44 A;

Figure 45 A shows and support to use the increment VM/PVS(increment G increment G-sub-grid that sends from remote server) exemplary process diagram of the data storage safeguarded of the increment VM/PVS of data and transmission method;

Figure 45 B is the continuation of Figure 45 A;

Figure 45 C is the continuation of Figure 45 B;

Figure 46 shows the block diagram/process flow diagram of distributed clients-server implementation, outer increment G sub-grid data of the nuclear stored of server access wherein are converted to DDL, VM/PVS data on DDL ID and maintenance server and the client with Global ID;

Figure 47 shows the block diagram/process flow diagram of following layout: server process is positioned on the identical equipment with client process and uses the single part of copy of DDL;

Figure 48 A shows the block diagram/process flow diagram that can penetrate also very soon in the situation of looking elementary boundary that does not read/send the increment visual information server process that sends the level of detail increment visibility information that has reduced in current view point;

Figure 48 B shows the exemplary process diagram that prevents the method that the visibility events message delay arrives by reducing the viewpoint speed that allows;

Figure 49 shows the exemplary diagram of the impact of exposing the time differentiates spatial detail on the human subject ability;

Figure 50 A shows in following situation the block diagram/process flow diagram of the server process that sends the level of detail increment visibility information that has reduced: the recently visible geometric figure corresponding with the increment visibility information and/or the image-space velocity of texture are high enough to so that reduce for the visual sensitivity of kinematic geometry figure and/or texture;

Figure 50 B shows the block diagram/process flow diagram of alternate embodiments that has surpassed the server process of the level of detail visibility events information that transmission in the situation of available transmission bandwidth reduced at required transmission bandwidth:;

Figure 51 shows the exemplary diagram as the relative spatial vision sensitivity of the function of retina speed;

Figure 52 shows the exemplary process diagram/block diagram of following client and server process: the looking the increment G sub-grid of unit and be configured to the observability figure that the father looks the unit/PVS with server access pre-stored of this information to(for) the father, and then it be used from the profile contour information one of looking the unit transition for son of pre-stored and be configured to the increment G sub-grid data that son is looked the unit transition;

Figure 53 shows the prior art of incomplete uncontinuity grid wedge and draws;

Figure 54 shows the exemplary process diagram/block diagram of following client and server process: client increment G data and mark profile contour data and use this data to generate PVS and/or increment PVS data;

Figure 55 determines and the illustrative diagram of the system and method that the content flow of increment PVS is processed for the observability of implementing to look from the district according to the basis of use Conservative Linear umbra event face of the present invention;

Figure 56 shows the exemplary diagram of processor; And

Figure 57 shows the exemplary diagram of the relation between visibility events scrambler in a kind of embodiment, visibility events server and the visibility events client.

Embodiment

In the exemplary embodiment, the effective static barriers of term ESO() refer to the quantity of polygon (original mesh) in the district that is blocked with observability figure and/or the tolerance that these polygonal surface areas become certain direct ratio.ESO also with around the new polygonal quantity of introducing in the visible range of distinguishing that is blocked becomes certain inverse ratio, and above-mentioned new polygon is introduced into by the again trigonometric ratio that the limit of blocking the border (edge) causes.This tolerance is used for the conservative simplification of VM or unified VM.

In the exemplary embodiment, term EDO(effectively dynamically blocks) refer to the increment district (DR) of blocking in polygonal quantity and/or polygonal area of being blocked become measuring of certain direct ratio, wherein DR is illustrated in and specifically looks the blocked area that the unit transition period produces.EDO also with visible range around DR in the new polygonal quantity introduced become certain inverse ratio, new polygon is introduced into by the again trigonometric ratio that the limit of blocking the border causes.

In the exemplary embodiment, the effective dynamically observability of term EDV() refer to the measurement of validity in the increment district (DR) of unified observability figure.If DR is for the DR that specifically looks the unit transition _O(the increment district of blocking), then EDV is corresponding to the EDO of DR.

If DR is DR _E(exposing the increment district), then by check around the ESO of blocked area determine EDV.By around OR simplification and with DR _EPolygon extend to OR or DR _OIn proceed DR _ESimplification.

In the exemplary embodiment, term is unified observability figure and is referred to comprise the observability figure that blocks the border that looks from looking the unit that looks unit (for example A and B) generation according to two, wherein looking the unit is associated with one of following dual mode: 1) look the unit and be completely contained in another and look in the unit for one, or 2) look the unit and fully share a common surface.The layout that unified observability figure is the VM district so that: for from looking unit A to the transition of looking unit B, some districts comprise the mesh triangles shape/segmentation that recently is blocked and other districts comprise the mesh triangles shape/segmentation of recently exposing.Unified observability figure is used for being configured to the directly increment PVS data of storage.Alternately, unified observability figure can be used for identification and block significantly or profile contour significantly, and it can be labeled and be used for subsequently generating increment G increment G/ increment PVS data.

In the exemplary embodiment, term wedge (also referring to CLUES) refers to by the formed visibility events face of the feature of looking the unit (summit or limit) and the polygonal summit of grid or limit.Usually, wedge defines from the feature of looking the unit and passes the polygonal summit of grid or the observability looked in the limit.

The wedge that adopts in the art methods of uncontinuity gridding is accurate.These limits can be plane or quadric surface.Plane wedge described in the discontinuous grid document is two classes of rename here:

1) SV-ME wedge-formed by the summit of looking unit (or " source ") and the limit of grid.Be also referred to as the rotation wedge or support the summit wedge.

2) SE-MV wedge-formed by the summit of the limit of looking the unit and polygonal mesh.Be also referred to as and scan wedge or support edge wedge.

3) SE-ME wedge-at the grid silhouette edge is parallel to support in the specific situation look the unit silhouette edge and forms.

These definition supposition orthogonal projection (that is, use is looked the unit as light source).In back-projection approaches, the segmentation of silhouette edge or silhouette edge is as the various silhouette edges support back projection event faces in " source " and limit, source and the axle of looking between the unit.In addition, definition is identical to the situation of back projection.

Because the wedge that adopts in the uncontinuity gridding is generally used for identifying the part of source penumbra, these parts are configured on relatively a large amount of limit of polygonal mesh, and above-mentioned limit is called the silhouette edge of looking from looking the unit.

Because employed plane wedge is accurate event face in discontinuous shape gridding, so they are not limited at inferior segment: the element characteristic (summit or limit) of looking of wedge is looked from the polygonal mesh feature of wedge and is blocked.This definition of wedge produces that to cause face be " gap " in the event face of discontinuous plane.In uncontinuity gridding method completely, it can be the high-order visibility events face of quadric surface wedge that these gap-fills have.The gap is filled by these high-order event faces, so result's visibility events face is normally continuous.

Referring to the Table I a and the Table I b that are the wedge nomenclature.

Embodiment also adopts the event face of looking from feature, similar but be different from the conservative linearization umbra event face (CLUES) of these wedges aspect important with the plane wedge that adopts on plane discontinuous network is formatted.

Employed plane wedge and CLUES(are also referred to as the single order wedge in this manual in the uncontinuity gridding, or be called simply wedge) between a difference be: the wedge that adopts in the method only is can form from looking those wedges of the umbra event face of looking the unit, and penumbra event itself is not considered from look the observability of looking the unit.The wedge of this method is configured and constructs them with the technology of scanning of the event wedge that only generates potential umbra at less polygonal mesh limit (silhouette edge that is called single order).This means, the quantity of the wedge of constructing in the method is much smaller than the format quantity of the middle wedge that generates of discontinuous network.

Another difference that discontinuous network is formatted between the wedge of wedge and this method is: wedge only uses the polygonal mesh feature of looking element characteristic and wedge of wedge to be defined and to construct.Any geometric figure between two parties between these two features will be left in the basket.

The first order modeling that this method of wedge structure is propagated based on the observability in the polyhedron environment, this first order modeling guarantees the border of conservative, the continuous umbra of structure.

In fact, between two parties geometric figure can produce look element characteristic from district that the polygonal mesh feature is looked and is blocked.These are the wherein not limited district of discontinuous grid wedge of correspondence (therefore producing gap or uncontinuity in the event face of usually being filled by high-order wedge or quadric surface) of wedge.By ignoring this geometric figure between two parties, the wedge that this method structure limits the gapped continuous event face of tool not.Owing to blocking the wedge of this type structure this method by ignoring high-order, wedge conservatively represents the event face of actual umbra of looking from feature.For the geometric district between two parties that do not have of wedge, the wedge of being constructed by this method is accurate.

Wedge is in coarse district therein, and is optional the, and wedge can be substituted by other wedges that the method for the modification of using the wedge structure is constructed, and the method for modification explains that the high-order that is caused by geometric figure between two parties blocks.

This method comprises three types (single order) wedge:

1) SV-ME wedge-form by extending corresponding rotary support polygon.The support polygon of corresponding rotation by the single order silhouette edge of the support summit of looking unit (SVV) and polygonal mesh by processing to form from the limit to the rotation of looking the unit.The rotary support polygon is also referred to as SV-ME support polygon or summit support polygon.(profile) limit reflection of such visibility events face by grid looked comprising of some place on the unit and blocked.Be also referred to as the wedge of rotation.This rotation is processed the processing that is described as identifying the single order silhouette edge and looks the supporting plane between the unit.Although In the view of the mankind process can be the plane about the continuous rotation of the reality of silhouette edge until touch and look the unit, in fact, embodiment can be measured by each candidate's supporting plane (by correspondence look the summit, unit forms) and another polygon formed specific discrete angle.Relatively these angle measurement make it possible to determine actual support polygon according to the quantity of candidate's support polygon in one embodiment.

2) SE-MV wedge-by the limit of scanning support polygon of correspondence is extended to form, (also be called simply and scan polygon or limit support polygon), the corresponding support polygon that scans is by the support edge of looking the unit and inside corner grid profile summit, by looking the formed support polygon of processing that scans unit profile contour (SVSC) along the support between the SVV that supports adjacent SV-ME wedge.Comprising on limited (border) limit of looking the unit located on such visibility events face reflection (inside corner) grid profile summit.The SE-MV wedge is also referred to as and scans wedge.

3) SE-ME wedge-is only looked element sides and is just formed when parallel with the grid silhouette edge that supports when supporting.By extending to form on the limit that corresponding SE-ME support polygon between element sides and the grid silhouette edge that supports is looked formed in parallel support.The plane wedge that is different from other types is determined the visibility problem of looking from the district for the observability about wedge of SE-ME wedge, rather than the visibility problem of looking from point.Comprising and block on (border) limit of unit looked in (profile) limit reflection of such visibility events face by grid.

In the uncontinuity gridding of the prior art among employed wedge and the present invention another the important difference between employed those wedges be: use in the method profile summit about wedge to be constructed to appear at single order and determine observability about wedge from the conservative method on the silhouette edge of looking the unit and looking.This guarantees that each profile summit about wedge is compound profile summit (CSV), the cross-shaped portion of two wedges (wedge is corresponding to current wedge).On the contrary, in the uncontinuity gridding method of prior art, usually use the object space observability method of looking from point such as Weiler-Atherton algorithm accurately to determine observability about wedge.

In the exemplary embodiment, term rotation wedge refers to the SV-ME wedge that the limit of the support polygon by extending rotation forms.

In the exemplary embodiment, the conservative linearization umbra event face of term CLUES() (referring to wedge) refers to use another title of the single order umbra wedge that rotation of the present invention and the method for scanning construct.Can use backprojection of the present invention these wedges of refinement mutual with reflection high-order observability.

In the exemplary embodiment, term umbra boundary polygon (UBP) refers to as the polygon from the part of the face of looking the umbra volume of looking the unit.In the method, can use the conservative UBP that derives from (single order) wedge of correspondence to construct the umbra volume (umbra that is called the polyhedron polymerization, or PAU) of looking the unit from looking.

The umbra event face of looking from looking element characteristic by the employed wedge of this method, its be guaranteed for next-door neighbour only support the grid silhouette edge of this wedge from looking the umbra event face (looking from the whole unit of looking) of looking the unit.This is because this wedge can be intersecting from looking the mode of umbra boundary limitation on wedge of looking the unit and another wedge that exceeds the support profiles limit.That is to say, look look visible wedge tangentially of element characteristic itself and can become from looking other parts of looking the unit as seen from what support.

High-order wedge according to correspondence can be constructed high-order UBP.

In the exemplary embodiment, the term polygonal mesh refers to the finite aggregate of summit, limit and the face (being also referred to as polygon) of the connection that is connected with the limit according to the summit.If two polygons of grid intersect, limit or the summit of intersecting must be the parts of grid.Allow not have the penetrating mutually of face (interpenetration).Be also referred to as polygonal mesh object, triangular mesh, perhaps be called as simply grid.If each limit of grid is by two polygons are shared at the most, then grid is stream shape polygonal mesh.If each limit is accurately shared by two faces, then grid is closed stream shape polygonal mesh.Unless otherwise, otherwise suppose that the polygonal mesh in this instructions is closed stream shape polygonal mesh.

In the exemplary embodiment, the polyhedron that unit or vision area refer to be expressed as polygonal mesh looked in term, and it has described the district that viewpoint is limited.Unless otherwise, otherwise suppose in this instructions look the unit and vision area is protruding.Be restricted to parallelepipedon or box depending on the unit, but can do restriction like this to vision area.

In the exemplary embodiment, the potential visible collection of term PVS() refer to from looking the look set of visible polygon or polygonal segmentation of unit.Usually PVS is calculated to be conservatively, comprises all visible polygons or polygon segmentation and some sightless polygons or polygon segmentation.

In the exemplary embodiment, term polyhedron polymerization umbra (PAU) refers to suppose that from looking the look volume in the space of being blocked by mesh object of unit the first order modeling that observability is propagated is called as single order polyhedron umbra volume.Because each umbra volume can intersect with polymerization and block, we claim that these volumes are single order polyhedron polymerization umbra (PAU).

Single order PAU, also referred to as PAU, the polygon that is called as umbra boundary polygon or UBP surrounds.These polygons intersect and form by having the polygonal single order wedge of triangular mesh and other wedges.PAU is also surrounded by the visible grid polygon of single order segmentation (comprising from looking the segmentation of the observability figure that looks the unit).UBP forms continuous (although need not to be sealing) the umbra face on the border that limits PAU with the segmentation of visible grid polygon.

As what describe in detail in conjunction with 3D2-stream shape traversal method (Figure 20 and relevant figure): the structure of observability figure relates to following step: determine to be positioned at observability ambiguity line segment about wedge and be actually the no PAU of being positioned at volume and also therefore look and be blocked from the whole unit of looking.The method comprises polyhedral some pardon (point-in-polyhedron) test of modification, and this test can be in the situation that there be clear and definite structure to construct whole PAU answer for this inquiry of single order PAU.

In the exemplary embodiment, term uncontinuity grid (DM) refers to intersect formed grid by visibility events face and grid polygon.According to being incident on the formed discontinuous grid of the visibility events face of looking on the unit grid polygon is divided into subregion (district) with respect to qualitatively observability or " aspect " of the unification of looking the unit.

In the art methods of completely uncontinuity gridding, all the event faces, umbra and the penumbra that are incident on the light source are constructed.

In some embodiments, according to single order from the umbra visibility events face looking the unit and look or according to by the single order umbra visibility events face of refinement, construct from looking the uncontinuity grid of looking the unit by the back projection technology and to interact to explain the high-order observability.

Although only adopt the fact of umbra event face; But all districts that are not the umbra DM that adjoins with the side that is blocked of oriented DM polyline in fact all look and are blocked from the whole unit of looking.This be because district's (it is actually the part of PAU) from look the state of looking the unit be by among the R3 may be not the wedge-wedge cross-shaped portion of reflection in the wedge of correspondence-polygonal mesh cross-shaped portion determine.

In the exemplary embodiment, term observability figure (VM) refers to the grid polygon is divided into from whole the look district that is blocked and from looking certain point on the unit other districts that look visible, unit of looking.In the art methods of the observability of looking from the district accurately people 2000,2005 such as () Nierenstein, construct these subregions with being generally quadric accurate visibility events face.

Embodiment uses the corresponding conservative linearization umbra uncontinuity grid of CLUES structure.Result's DM is divided into the grid polygon from whole the look district that is blocked and from looking the look conservative division in visible other districts of certain point on the unit, unit of looking.Because all districts that are not the umbra DM that adjoins with the side that is blocked of oriented DM polyline in fact all look and are blocked from the whole unit of looking, so the border of VM is the subset on the border of corresponding DM.By contrast, corresponding VM only comprise be guaranteed for from whole look the unit look (this shadow zone of VM) be blocked and from looking certain point on the unit other districts that look visible, wherein block and may conservatively be underestimated, thereby observability over-evaluated.

In the exemplary embodiment, the term silhouette edge refers to have the limit of assembly polygon just facing with respect to ad-hoc location and another the assembly polygonal polygonal mesh right with respect to the same position back side.

In the exemplary embodiment, the term silhouette edge of looking from point refers to have the limit of assembly polygon just facing with respect to specific point and another the assembly polygonal polygonal mesh right with respect to the same point back side.

In the exemplary embodiment, the term silhouette edge (being also referred to as the general silhouette edge of looking from the district) of looking from the district with respect to as the district of looking unit (or the polygonal mesh limit in the back projection situation) that is used for light source limit.If the position is for looking the unit, the silhouette edge of then looking from the district can be called as the silhouette edge of looking from looking the unit.If this Qu Weibian, the silhouette edge of looking from the district so can be called as the silhouette edge of looking from the limit.In this manual, the silhouette edge of any type (look, look, look from the limit from looking the unit from point) can be called as silhouette edge simply, and the type of silhouette edge is suggested by context.

From look general the silhouette edge of looking the unit be as for look arbitrfary point on the unit (or area source) from the look any limit of polygonal mesh of silhouette edge of point.This be by people such as Nierenstein 2005 that adopt and in the people such as Drettakis 1994 completely uncontinuity gridding method from looking the definition of the silhouette edge of looking the unit.

Usually, the penumbra event face of looking from the district is supported on these limits, but in fact subset supports quadric umbra event face of looking from the district that is generally.

When the high-order observability of considering the limit tlv triple is mutual, can accurately limit the silhouette edge of looking from the district.Alternately, as in the method, by only be considered as the limit between the visibility events face that occurs of crossing result, can conservatively limit the silhouette edge of looking from the district; Single order observability model as the observability propagation.

In the exemplary embodiment, what term single order silhouette edge referred to single order is the limit that has with respect to the whole polygonal polygonal mesh of another assembly of looking right assembly polygon of the back side, unit and just facing with respect at least one summit of looking the unit from looking the silhouette edge (also being called simply the single order silhouette edge) of looking the unit, wherein the assembly polygon back side pair relative to each other.

This definition is based on simple, the conservative pattern that is called the single order observability that the observability in the polyhedron environment is propagated, this pattern only be considered as the limit between crossing result's visibility events face.

One embodiment of the present invention are adopted as the polygonal mesh of stream shape triangular mesh.In stream shape triangular mesh, each limit is accurately fully shared by two triangles.By simplify the explanation of single order silhouette edge with stream shape triangular mesh.

The single order silhouette edge with respect to looking the unit of polygonal mesh be polygonal mesh with respect to the local support limit of looking the unit.If only consider to look two assembly polygons (triangle) on unit and shared limit in supporting test, then the polygon of looking between unit and the limit is supported on the local support limit.(referring to the definition of supporting test)

Usually, the single order silhouette edge of looking from the district is the little subset of the accurate silhouette edge of looking from the district of arbitrary polygon grid.

In this manual, appointing of the single order silhouette edge of any type (look, look from the limit from looking the unit) can be called the single order silhouette edge simply, or is called simply silhouette edge, and the type of silhouette edge is hinted by context.

The present invention includes a kind of identification (by adaptive refinement and back projection) single order silhouette edge where and be coarse and with the method than near side (ns) of silhouette edge " retraction " to the set of the accurate silhouette edge of looking from the district that belongs to polygonal mesh.

In the exemplary embodiment, about term local support limit, referring to the single order silhouette edge.

In the exemplary embodiment, the term support polygon refers to the support polygon by two kinds of structures " support ".In the method, in one case, the single order silhouette edge of polygonal mesh and the support polygon of looking between the unit are to form (SV-ME support polygon) by this single order silhouette edge and this summit of looking the unit.Be called as support depending on the polygonal summit of unit supports and look summit, unit (SVV).Can identify support by the polygonal plane of back side assembly on rotation profile limit and look the summit, unit, wherein, rotation around silhouette edge occur and towards looking the unit along the polygonal normal direction of back side assembly on limit until the plane of support polygon with look the unit and intersect.In the ordinary course of things, this intersects to occur in to support looks place, summit, unit, and it forms leg-of-mutton support polygon together with the single order silhouette edge.Look the summit that the summit, unit is the limit of looking unit parallel with the silhouette edge of grid if support, Plane of rotation will intersect with the limit of looking the unit so, be not single summit, and support polygon will be by grid silhouette edge and the quadrilateral that element sides forms of looking that intersects.This Equations of The Second Kind support polygon is called as the SE-ME support polygon.

In the another kind of situation of this method, between the inside corner summit of single order silhouette edge and the limit of looking unit (the SE-MV support polygon is also referred to as support triangle shape), form the support polygon of different types.

In the context of the present invention, support polygon conservatively is defined as because single order silhouette edge (being also referred to as the local support limit) or its summit and the corresponding unit supports of looking, ignores the single order silhouette edge and look any between the unit and block or disturb.If for example intersected by the defined support polygon of the present invention and single order limit and the geometric figure looked between the unit, then support polygon be not such as in the prior art definition support polygon (if interference exist, then it generally can not limit support polygon).

As defined in the prior art, if polygon is by the summit of a structure or limit and do not support with other summit or the limit of anything another structure that intersects, polygon will be by " supporting test " (that is, being support polygon) between two structures so.Supporting test also requires: support polygon along away from first support object (as, depending on the unit) direction extension (for example, this expansion is " wedge ") so that support polygon be in another support structure inside (for example, " inside " of the topology of stream shape grid) mode does not intersect with another structure (for example, polygonal mesh).This support test effectively require the support edge be structure (for example, polygonal mesh) " outside " limit, this structure supports to structure tangentially with support polygon, with isostructural " inside " limit of polygonal mesh or the reflection limit as support polygon not being supported to tangentially structure.

In the method, support the polygon of test by only comprising the limit that shares grid, be used for determining in more limited mode whether the limit supports the conservative support polygon looked between unit and the grid (namely, whether the limit is " local support " or single order silhouette edge, referring to the definition of single order silhouette edge and local support edge).

In the situation of emphasizing the difference that the prior art definition of the definition that the basis of support polygon is conservative with support polygon separates, the support polygon of the present invention's definition can be called conservative support polygon.Otherwise the conservative support polygon that will define in the present invention is called support polygon simply.

As defined in the present invention, the wedge that derives from (conservative) support polygon always forms the continuous conservative linearization umbra event face that can intersect with the grid polygon conservatively determining from looking the unit set of visible grid polygon (or its segmentation) of looking, and does not need the accurate quadric surface of separating of common domination (and complicated).

In the exemplary embodiment, about the conservative support polygon of term, referring to above-mentioned term for support edge shape.

In the exemplary embodiment, term support that test refers to if polygon by the summit of a structure or limit and do not support with other summit or the limit of anything another structure that intersects, then will be by the polygon of " supporting test " (that is, being support polygon) between two polygonal structures.Supporting test also requires: support polygon along away from the first object that supports (as, depending on the unit) direction extension (for example, this expansion is " wedge ") so that support polygon be in another support structure inside (for example, " inside " of the topology of stream shape grid) mode does not intersect with another structure (for example, polygonal mesh).This support test effectively require the support edge be structure (for example, polygonal mesh) " outside " limit, this structure supports to structure tangentially with support polygon, with isostructural " inside " limit of polygonal mesh or the reflection limit as support polygon not being supported to tangentially structure.

In the method, support the polygon of test by only comprising the limit of sharing grid, be used for determining in more limited mode whether the limit supports the conservative support polygon looked between unit and the grid (namely, whether the limit is " local support " or single order silhouette edge, referring to the definition of single order silhouette edge and local support edge).

In the situation of emphasizing the difference that the prior art definition of the definition that the basis of support polygon is conservative with support polygon separates, the support polygon of the present invention's definition can be called conservative support polygon.Otherwise the conservative support polygon that will define in the present invention is called support polygon simply

In the exemplary embodiment, the conservative support package of term refers to by the formed polygonized structure of conservative support polygon between a polyhedron (for example, depending on the unit) and one or more other polyhedrons (for example, polygonal mesh object).Rotation and the method for scanning are looked the method for the specific subset of the conservative support package between unit and the non-convex polygon mesh object for structure.

Support package is the summary of " convex closure ", and " convex closure " is prior art important in computational geometry image and linear programming.Convex closure between two convex polyhedrons is polygonized structure, comprises all " sight line " of two observabilities between the convex polyhedron.The art methods that forms the convex closure between convex polyhedron (for example, depending on the unit) and another convex polyhedron (for example, convex polygon grid) is known and important.These art methods adopt the structure of the support polygon between two convex objects.(referring to See O ' Rourke, Computational Geometry in C Second edition Cambridge University Press1998).

Not about the description of the obvious prior art of formation support package between convex polyhedron and one or more non-convex polyhedron (the polygonal mesh object that for example uses among the present invention, and they are ubiquitous in computer graphical).Accurate support package will not only comprise polygon, also comprise the quadric surface that is incident on the compound profile summit.

On the contrary, use the set of the conservative support polygon that rotation of the present invention and the method for scanning can construct to be added at an easy rate (being incident on the wedge that scans on the outside angular vertex of polygonal mesh by interpolation), continuous to form, the conservative accurate support package that is similar between convex polyhedron (for example, depending on the unit) and one or more non-convex polyhedron.

In one embodiment of the invention specified rotation and scan the conservative polygonal subset of support package of method construct, when conservative support package polygon extends, can be used for determining not need quadric conservative continuous umbra event face from looking the unit look visible polygonal set in conjunction with forming, form wedge.

Be configured being included in some polygons in the fully conservative support package not in a kind of rotation of embodiment and scanning in the method, because corresponding wedge (for example, scan, or the outside of inciding polygonal mesh is than the SE-MV wedge on the angular vertex) what does not help continuous umbra border be that what looks visible be from looking separating that the unit looks and be blocked with from looking the unit with.

Rotating and scanning in the method, do not identify these support polygons.Therefore, do not construct the wedge of their correspondence.

Except specified rotation with scan the method, it is feasible adopting the alternative embodiment of the continuous umbra event face of conservative support polygon structure.For example, alternative embodiment can be configured in looks the whole fully conservative support package of unit between the polygonal mesh object, then wedge is extended to form in the polygonal limit of all support packages.The wedge that so forms comprises the wedge (for example, by extending formed wedge by the support polygon that the lateral angle summit of the limit of looking the unit and polygonal mesh is supported) of not facilitating continuous umbra event face.In so alternative embodiment, can ignore or remove these unnecessary wedges.

In the exemplary embodiment, term SVV(supports and to look the summit, unit) refer to, for given grid silhouette edge, what run into when passing the grid silhouette edge with the polygonal normal direction rotation of the back side assembly of silhouette edge first looks the summit, unit.(also referring to support polygon)

In the exemplary embodiment, term supports and to look unit profile contour (SVSC) and refer to a part of looking unit profile contour when observed from the inside corner summit of grid silhouette edge, this most extreme umbra border of part generation.This is the part of blocking from looking the generation minimum of looking unit profile contour when the unit profile is watched by inside corner grid profile summit.It still produces the SE_MV wedge that has the orientation consistent with the SV-ME wedge that connects and form continuous face when scanning operation.Support is looked unit profile contour and is extended between two SVVS corresponding with the grid silhouette edge that produces the inside corner summit.

The SE-MV wedge is the visibility events face of following orientation: by looking the limit visibility that reflects place, grid profile summit comprising on the elemental area.

On the contrary, the SV-ME wedge is the visibility events face of following orientation: (the looking from looking the unit) that causes by means of the grid polygon by the silhouette edge place blocks to reflect the limit visibility of grid silhouette edge.

SVSC looks the unit silhouette edge for (the looking from the grid silhouette edge) that produce the corresponding SE-MV wedge with orientation consistent with the orientation of adjacent SV-ME wedge; Thereby produce continuous, the conservative as one man umbra event face of orientation at place, grid profile summit.

In the exemplary embodiment, scan triangle referring to scanning polygon about term.

In the exemplary embodiment, term scans the visibility events border, non-protruding (or " inside ") angle place of single order silhouette edge that polygon (also referred to as scanning support polygon or scanning triangle) refers to polygonal mesh not only by being extended by those support polygons that the silhouette edge that forms this inside corner supports, also likely by the inside corner summit as at the grid profile and conduct, come and some limit of looking unit of the silhouette edge of looking from point of the distant view on the inside corner profile summit of mesh object between one or more polygons that scan of formed dissimilar support polygon form.Between the SVV corresponding with the inside corner of mesh object, form profile chain (unit profile contour is looked in extreme value or support) depending on these of unit from the some silhouette edge of looking.Polygon (triangle) " is scanned " each limit that goes out this chain, forms to scan polygon.These scan polygonal limit extend to form the inside corner place that also helps grid profile contour single order visibility events face SE-MV or scan wedge.

In the exemplary embodiment, term scans wedge and refers to by the formed SE-MV wedge of the extension on the limit of scanning support polygon.

In the exemplary embodiment, term separation polygon refers to separate the polygon of two kinds of structures.Generally speaking, the silhouette edge of polygonal mesh is formed by silhouette edge and the summit of looking the unit with the polygon that separates of looking between the unit.Supporting this polygonal summit of looking the unit is called as to separate and looks the summit, unit.Be rotated to identify to separate by the polygonal plane of back side assembly with silhouette edge and look the summit, unit, wherein, around silhouette edge and along the opposite direction of the polygonal normal of back side assembly on limit towards looking the unit rotates until the plane with look the unit and intersect.Generally speaking, this intersects will occur in to separate looks place, summit, unit, forms as the leg-of-mutton polygon that separates with silhouette edge.If it is the summit of looking unit parallel with the silhouette edge of grid that the summit, unit is looked in separation, Plane of rotation will intersect with the limit of looking the unit so, not only intersect with single summit, and the separation polygon will be by grid silhouette edge and the quadrilateral that element sides forms of looking that intersects.Be used for determining to be incident on the single order UBP of silhouette edge and the maximum deviation between the high-order UBP.

In the exemplary embodiment, term umbra boundary polygon (UBP) refers to look the unit by the polygon of the part on the formed umbra of polygonal mesh object border as using as area source.UBP can or can conservatively be similar to umbra border in the district corresponding to accurate umbra border.With rotation of the present invention with scan building method by support polygon and scan polygonal expansion and construct.Between tectonic epochs in the early stage, UBP extends semiinfinite away from looking the unit.In the follow-up step of structure PAU, UBP intersects each other, with the grid polygon and may intersect with the bounding box that surrounds all mesh objects.

In the exemplary embodiment, term single order UBP refers to use rotation and scans method and polygon that the embodiment that substitutes of the method described is in this manual constructed.

In the exemplary embodiment, summit, term single order SV-ME UBP(source-Grid Edge UBP) refer to by the grid silhouette edge is extended the polygon of constructing with the corresponding support polygon (SV-ME support polygon) of looking between the summit, unit.

In the exemplary embodiment, limit, term single order SE-MV UBP(source grid vertex UBP) refer to by the polygon of constructing is extended with the corresponding polygon (SE-M V scans polygon) that scans of looking between the summit, unit in limit, grid profile contour inside corner summit (simple or compound profile contour).

In the exemplary embodiment, about term SV-ME support polygon, referring to SV-MEUBP.

In the exemplary embodiment, term SE-MV scans polygon, referring to SE-MV UBP.

In the exemplary embodiment, term high-order UBP refers to use the UBP that the high-order model of the observability propagation in the polyhedron environment is constructed.This model solves part, the part of the light source (for example, depending on the unit) that can be blocked from the silhouette edge that exposes.In fact high-order UBP can be similar to wherein the umbra border by the reality in the district on the formed umbra of mutual formed high-order (secondary) the face border of Bian-Bian-Bian (EEE) more accurately.In the method, use back-projection approaches structure high-order UBP.

High-order UBP can incide on the single order silhouette edge, and wherein high-order UBP is called as adjusting UBP.Alternately, high-order UBP can incide on the high-order silhouette edge.Violate local observability if regulate UBP, can calculate the high-order silhouette edge.

In the exemplary embodiment, the term back projection refer to look unit (light source) from silhouette edge the determining of part of looking visible.In the method, this definite employing is used silhouette edge as the rotation of the PAU structure of light source and is scanned method.

In the exemplary embodiment, the visible support of term VSVV(looked the summit, unit) refer to for grid silhouette edge or the determined summit of limit part: the visible support of in fact looking from the limit is looked the summit, unit and is determined by the method for back projection.Be used for the SV-ME UBP that structure is adjusted.

In the exemplary embodiment, the visible extreme value of term is looked the unit profile visible extreme value that refers in fact look from the inside corner summit of grid profile and is looked the unit profile.Be used for the polygon that scans that structure is extended to form high-order SE-MV UBP.

In the exemplary embodiment, the simple profile contour of term refers to the profile side chain by the summit that the shares connection that belongs to single mesh object.Be also referred to as simple profile.

In the exemplary embodiment, term compound-contoured contour refers to comprise the profile side chain of the silhouette edge that is connected by the summit of sharing or be connected with the crossing formed summit of non-adjacent silhouette edge by the wedge/UBP from a profile.In the research of smooth manifold, such cross-shaped portion is called as the t joint.(referring to See Durand, Fredo PhD thesis University of Grenoble)

In the exemplary embodiment, about term t joint, be also referred to as compound profile summit (CSV), referring to compound profile contour.

In the exemplary embodiment, term PAU(polyhedron polymerization umbra) refer to form used by one or more polygonal mesh objects and look the unit as the polyhedron on the border of the umbra of light source projects.PAU is expressed as the polygonal mesh of the visible segmentation that comprises UBP and polygonal mesh object.

In the exemplary embodiment, term " triangle-VC axle " (triangle-look unit shaft) refers to mesh triangles shape and the protruding axle (back shaft or convex closure) of looking between the unit.

In the exemplary embodiment, term " segmentation-SILE axle " refers to the 2D axle between MSEGMENT and the silhouette edge.Be used for the 2D version of trellis traversal to find the polygonal cross-shaped portion of UBP and grid.

In the exemplary embodiment, term UBL(umbra boundary line) refer to that the 2D of UBP is equivalent, this 2D equivalence is formed between silhouette edge summit and the grid profile summit.

In the exemplary embodiment, term PLAU(multi-line section polymerization umbra) refer to that the 2D of PAU is equivalent, this 2D equivalence is limited to the face of UBP.

In the exemplary embodiment, term is looked unit profile contour and is referred to the profile contour of looking the unit as observing from some elements of triangular mesh.

In the exemplary embodiment, polygonal mesh can be represented as digraph.In the exemplary embodiment, the term trellis traversal refers to: such graph traversal is the process of the node of access figure.In the exemplary embodiment, trellis traversal can be followed and wherein check the adjacent polygonal breadth-first order in limit.Other traversal sequential are possible.

In the exemplary embodiment, the summit, unit is looked in support about term, referring to SVV.

In the exemplary embodiment, the term support is looked element sides (SVE) and is referred to the limit of looking unit parallel with corresponding grid silhouette edge.Support polygon between two limits is quadrilateral.

In the exemplary embodiment, the visible support of term looked element sides (VSVE) and referred to from the part of the silhouette edge of whole correspondence visible (not blocking) SVE.

In the exemplary embodiment, the profile contour that blocks significantly of term SOSC() for looking the unit transition.

In the exemplary embodiment, term SESC refers to for the profile contour that exposes significantly of looking cell translation.

In the exemplary embodiment, the profile contour of term stream shape grid refers to observability and the fold singularity (fold singularity) of the stream shape that the transition phase between blocking is corresponding.For polyhedron stream shape grid, the profile contour is the polyline of piecewise linearity.

In the exemplary embodiment, the term tip refers to represent the some singularity of profile contour of the terminal point of profile contour.Non-protruding stream shape grid can have a plurality of tips, and each is corresponding to the terminal point of profile contour.

In the exemplary embodiment, term CSV(compound-contoured summit) refer to the phase cross-shaped portion of wedge and silhouette edge.Realize for single order, wedge is that single order wedge and silhouette edge are the single order silhouette edges.From the topological structure aspect, CSV is corresponding to the t-summit of the conservative definition of the profile contour of looking compound from the district.Usually, the inside corner of compound grid profile contour appears at the CSV place.

A kind of CSV of more rare type can appear in theory the place that wedge and profile contour intersect and go out.The situation of this degeneration can be corresponding to the external corners of compound-contoured contour.

Corresponding to the t-summit.

In the exemplary embodiment, the term wedge refers to formed triangle (SE-MV wedge) between the support summit of light source/look unit and silhouette edge.When the limit of silhouette edge and light source/look unit was parallel, wedge was in silhouette edge and supporting light sources/look between the unit forming.In this case, (SE-ME) wedge is quadrilateral.

Employed wedge is not limited on the part of blocking between source and the profile in discontinuous gridding method.Such wedge generation is accurate but might not produces the plane visibility events face on continuous umbra border.

On the contrary, the single order wedge is defined as whole support triangle shape or quadrilateral in the extension of looking between unit and the silhouette edge.The generation of single order wedge is accurately or guards, but always produces the plane visibility events face on continuous umbra border.

In another embodiment, wedge is any desired polygon of looking between unit and the polygonal mesh.

Wedge and UBP(umbra boundary polygon) different being: the scope of wedge only is subjected to and the polygonal crossing restriction of grid.The structure of UBP is not only by determining with polygonal intersecting of grid, also by determining with intersecting of other UBP.In fact, intersect and intersect to form the wedge of UBP with the grid polygon according to the wedge with other, form UBP.Limit the umbra border and be subset for the wedge of stream shape for the combination of the UBP of stream shape.

Can construct PAU with wedge-wedge and the crossing UBP that forms of wedge-grid polygon by direct.In this case, test to determine the geometric figure of PAU inboard with polyhedral some pardon.

Alternately, in the situation that there is not wedge-wedge to intersect, travels through by the visible sideline of only wedge-polygon being intersected and indirectly to construct PAU.In this case, penetrate the geometric figure that PAU inside is determined in test with geometric figure and the wedge of looking the line between the face of unit.

In the exemplary embodiment, the model that the observability that term single order observability (be also referred to as observability propagate first order modeling) refers to look from the district is propagated, (for example wherein suppose whole vision area, depending on the unit) be to look visible from the single order silhouette edge, look from the district that umbra event face incides that (single order) is visible, on the single order silhouette edge and be configured (use rotation and scan method).

In the exemplary embodiment, term high-order observability refers to the model that observability is propagated, and this observability is propagated and do not supposed that or not is visible that whole vision area (for example, depending on the unit) is looked from the limit of model.If support corresponding to single order silhouette edge (for example, SVV or SVE) and to look unity element and look invisible from the single order limit, so corresponding single order event face is coarse.In this case, the more accurate event face of looking from the district can be configured by back projection: use the single order silhouette edge as the source, and unity element (summit or limit) is looked in definite corresponding visible support.Back projection is processed can adopt first order modeling, also can adopt voluntarily high-order observability (by searching the SVV of source silhouette edge).Process by segmenting coarse single order limit and allowing alternatively the profile contour to cancel back projection, produce the umbra event face of assembling at accurate secondary event face in limit mode.

In the exemplary embodiment, the term back side refers to polygonal orientation.The polygon of orientation has a front and a back side.Each polygon is included in the plane that also has corresponding front and back.If polygon is with respect to backward the facing of point, this is at the back side on polygonal plane so.Determine whether polygon adopts polygonal plane equation with respect to a test of backward of point.

Determined the orientation on plane by normal vector, normal vector is limited by coefficient A, B and the C of plane equation:

Ax+By+Cz+D=0

Satisfy following inequality such as fruit dot, point (xp, yp, zp), then this is at the back side on this plane:

A（xp）+B（yp）+C（zp）<0

Otherwise this point is in the plane or in the front on plane.

Polygon also can be directed with respect to another polygon.If two polygons are shared limits, a kind of method of determining so their orientation is to select not to be to share the summit of polygon 2 on the summit on limit.Next, determine selected summit whether at the back side on the plane of polygon 1, in this case, two polygons are the back side pair relative to each other, otherwise they are relative to each other just in the face of (or at grade).Realize above object and advantages according to present embodiment, with and other objects and advantages, present embodiment comprises the method for observability precomputation conservative, that look from the district, wherein from the stereoscopic unit of multiaspect potentially visible polygon segmentation conservative by constructing, linearizing, determine from looking unit observability figure.

In one embodiment, mesh object forms by the triangular mesh of the stream shape of closure (wherein each limit is shared by two triangles definitely), although embodiment can use other polygonal mesh.The method can also be held non-closed jet shape polygon/triangular mesh that each limit is wherein shared by one or two triangle.

Use comprises conservative from looking conservative the linearization umbra event face (CLUES) on umbra border, unit, constructs conservative observability figure according to mesh triangles shape.

In this instructions, be also referred to as the single order wedge or simply the CLUES of wedge for relevant with the wedge that in discontinuous gridding method, adopts from feature visibility events face; Although they are different from discontinuous grid wedge aspect important.

Use is called the simplified model of the novelty that the observability in the polyhedron environment of single order observability propagates, and constructs CLUES on the triangular mesh specific limit of (being called the single order silhouette edge) (with the summit on these limits).The present invention includes the method for structure single order CLUES and adaptively with single order CLUES refinement to produce the method for more conservative linearization umbra event face.CLUES after these refinements reflection is looked the high-order observability impact that the unit causes when invisible from the support profiles limit when whole.The conservative polygon facet that use is easy to adopt, the linear event face after these high-order refinements often conservatively is similar to accurately (normally secondary) umbra border.According to some embodiments, when between single order event face and the high-order event face maximum deviation when surpassing predetermined value, carry out the refinement of single order event face.

In some embodiments, the silhouette edge by wherein supporting CLUES carries out thinning processing to determine that looking the unit as seen looks from the limit as the back projection of linear light source.

The first order modeling that observability is propagated is based on simplifying conservative hypothesis: if as seen silhouette edge looks from looking the unit, as seen silhouette edge from looking all parts of looking the unit so.For one of them assembly triangle with respect to look all on the unit the some back side to and another assembly triangle with respect to look on the unit at least one the point just in the face of and the said modules triangle do not face each other, this hypothesis causes the simple definition of single order silhouette edge.This definition is in fact identical with the definition of the silhouette edge of looking from point and reflected the following fact: some important aspect, in fact first order modeling will look the unit as point.

Simply the be rotated in single order silhouette edge structure of use from the limit to the strong point of looking the unit is called source summit-Grid Edge, or one type CLUES of SV-ME wedge.These SV-ME CLUES are similar to employed umbra boundary polygon of looking from point in shadow algorithm.Different from the umbra boundary polygon of looking from point, independent SV-ME CLUES not necessarily forms continuous umbra boundary surface non-protruding stream shape.

In the single order method, the CLUES(of Second Type is called source limit-grid vertex, or the SE-MV wedge) be configured, this CLUES is connected to continuous umbra event face with above-mentioned SV-ME wedge (by rotary configured).By scanning operation, the place, inside corner grid profile summit that does not form in addition continuous umbra event face from the wedge of the SV-ME type of adjacent silhouette edge therein forms the CLUES of this Second Type.Under these circumstances, the SV-ME wedge that incides on the adjacent single order silhouette edge connects to form continuous umbra event face by inciding the SV-ME wedge that connects inside corner grid profile summit.

According to by being anchored on scanning operation and run through when observing from inside corner grid profile summit as the formed support polygon that scans on the limit of looking the unit of silhouette edge and constructing the SE-MV wedge of inside corner grid profile summit place.Inside corner grid profile summit can be the summit by the formed simple profile of single order silhouette edge that connects.Alternately, inside corner grid contour summit can be formed compound-contoured summit (CSV) when single order wedge and another silhouette edge intersect.These summits are corresponding to the t-summit of the visible stream shape of looking from the district and when shape solution was looked visible in use accurately from the district, these summits were usually corresponding to the secondary event face.By at CSV structure SE-MV wedge, continuous, that guard, the linearization of blocking fusion that this method guarantees to be created on the inherence of not using the compound profile contour of reflection in the quadric situation umbra event face of looking from the district.

Table I b shows the visibility events face that also is shown in the Four types that adopts in the method that the complete discontinuous network among the Table I a formats.Visibility events face, CLUES in Table I b of the present invention also be presented and with complete discontinuous grid in the visibility events face that adopts compare.It should be noted that, the Jenkins nomenclature does not comprise secondary (EEE) face, because in observability propagation model of the present invention,, these quadric surfaces substitute by the SV-ME in the single order version of this method and SE-MV plane and with the back projection SV-ME/SE-MV of the high-order refinement embodiment in the method.

Table I b

The nomenclature of the visibility events face of looking from the district

In one embodiment, use CLUES to adopt the art methods of uncontinuity cancellated structure from the structure of looking the observability figure that looks the unit, wherein CLUES is replaced by the linearity that is used for the uncontinuity gridding and secondary " wedge ".Because it is responsive that the uncontinuity gridding method is not output, so this embodiment is not best.In the art methods of uncontinuity gridding, generate the event face at all silhouette edge, even the many silhouette edges in these silhouette edges may be looked and are blocked from being looked the unit.In the method, after having constructed all uncontinuity grid regions, determine the observability of uncontinuity grid regions.For the intensive environment that blocks, the many districts in these districts that construct look and fully are blocked from looking the unit.Therefore, the complexity of the layout of uncontinuity grid regions is far above the complexity (it is corresponding to observability figure) of the visible part of uncontinuity grid.

In another embodiment, CLUES be used for structure actual from looking the umbra volume of looking the unit, this umbra volume is called as polyhedron polymerization umbra (PAU), the mesh triangles shape segmentation that it comprises CLUES and is not blocked.The purpose of the method is only to determine the mesh triangles shape segmentation (it comprises potential visible collection or PVS) that is not blocked.The structure of whole PAU (it needs the potential cross-shaped portion of all CLUES) is usually for determining that the triangular mesh segmentation that is not blocked is optional.Alternatively, can come by the observability figure that direct structure is looked from the district more effectively to determine the mesh triangles shape segmentation that is not blocked.

Therefore, the present invention includes the output sensitive method of Conservative Linear observability figure structure, it is based on the traversal of triangular mesh 2-stream shape (being embedded among the R3).The breadth first traversal of the triangular mesh stream shape that is not blocked in the method.Jump to nearer untreated triangle if there is any triangle that blocks potentially (at triangle and look those triangles in the axle between the unit) also not to be traversed and to travel through, then end leg-of-mutton traversal.The method is carried out vertical order.Stream shape trellis traversal method marches to the profile contour of grid or from looking the border of blocking that the unit looks.The cataclysmal visibility events curve that is used as stream shape from looking the profile contour of looking the unit is processed.At these contour places, the CLUES that structure is corresponding and it is projected in the environment with that determine them and cross-shaped portion mesh triangles shape.This projection self is the visibility problem about face (for example, about wedge) that runs in the uncontinuity gridding, and previous the use is not responsive it being solved such as conventional arts such as Weiler-Atherton algorithms of output.Alternately, existing method comprises following technology: use the simple version that is suitable for the 1-stream shape that runs in about wedge or the visibility problem about CLUES of 2-stream shape traversal (now described) to solve visibility problem about wedge.For for simplicity, although it is different in importance to show the wedge of constructing for the single order wedge of this method and the art methods of formatting at discontinuous network, this instructions often uses word wedge, single order wedge and CLUES interchangeably.

Umbra event face and stream shape mesh triangles shape about the visible cross-shaped portion of wedge corresponding to from the segmentation of looking the umbra uncontinuity grid of looking the unit but may in fact not be the corresponding segmentation of blocking the border from looking the observability figure that looks the unit.This is because wedge represents from the observability of the triangle segmentation of looking the specific features (summit or limit) of looking the unit, not necessarily is from the whole umbra border of looking the unit of looking.In a kind of embodiment of this method, segmentation tests to be to determine that it is whether from looking the border of blocking that the unit looks it is constructed to umbra uncontinuity grid.

Block the border by carrying out vertical processing order and constructing when running into them, trellis traversal has been avoided to a great extent to the leg-of-mutton traversal that is blocked, and therefore is easy to realize the output sensitive property.In the output sensitive method, stream shape travels through effectively cascade and is over and flow to corresponding with the visibility events face profile contour that blocks on other stream shapes of intersecting on the border.

Result's output sensitive property is that following cost realizes: must be for untreated, the element that blocks potentially with triangle-looking unit shaft comes each grid elements that is not blocked is tested.In the method, by with level space segmentation with axle and these levels are comprised structure intersect and reduced widely the cost that these axles comprise test.All axles that the person has caused tending to O (NLog (M)) comprise the overall cost of test, and wherein N is the quantity of the visible grid elements that travels through, and M is the par of the grid elements of blocking potentially.

Come the grid polygon is processed by begin and continue the following arbitrary place of traversal in segmentation place of strong visible polygon: the section start or the b that a) are in the Conservative Linear umbra border wedge at silhouette edge place) wedge (forming actual in looking the border of blocking that the unit looks) and the polygonal intersection of grid.In order to ensure suitable depth order, any untreated grid elements of blocking potentially of trellis traversal algorithm identified and trellis traversal transferred to the nearer element that is not traversed immediately.Digraph by keeping expression triangle overlapping relationship and identify circulation among this figure with the algorithm of linear time algorithm such as Tarjan and detect uncertain depth order between the grid elements.When circulation existed, offending leg-of-mutton triangle and this axle of looking in unit-triangular shaft intersected to identify these overlapping leg-of-mutton parts that fully is arranged in this axle.These parts and coarse triangle can not form circulation, because they fully are positioned at this axle.Restart traversal with these parts and broken this circulation.

By carrying out the vertical traversal of grid, block boundary termination traversal and adopting the segmentation of level space, this algorithm is designed to realize exporting sensitive property, even for the environment that blocks thick and fast.

An advantage of trellis traversal/observability figure building method is: such as volume observability, generalized projection and Weiler-Atherton scheduling algorithm, this building method is more effective for identification is blocked.All these additive methods depend on uncommon large protruding shelter in real model.For example, only when the polygon that connects forms convex polyhedron, polygonal the blocking (they are called the processing of consolidation) that can be connected as the Weiler-Atherton algorithm of the observability algorithm of looking from point.Similarly, volume observability (people 2000 such as Schuaffler) method depends on to be formed on and looks unit and the simple axle between the single protruding box-shaped restraining mass of the shelter inside of reality.If actual shelter be recessed and or have the hole, the recessed restraining mass that then has been difficult to identify such simplification, it accurately represents blocking of actual shelter.

By contrast, the present invention does not rely on the existence of protruding shelter, but directly utilizes the connectivity intrinsic continuity of blocking of stream shape grid, and the shape of considering gridding not.

This method comprises " effectively static barriers " technology (ESO) in the district that is blocked that determines observability figure.Effectively blocking of district is to be reflected in the polygonal quantity that is blocked in the blocked area and the ratio of area, the additional geometric figure separation that the blocked area is produced during the caused again gridding in this district.

The precision of the observability figure that is produced by this method can reduce by the conservative protruding simplification of employed profile contour.This when around the observability map-area that is blocked block that the border comprises too many details the time can be useful, especially when effectively blocking of this district is very low.Effectively static barriers be used as controlling the profile contour simplification therefore and the inspiration of the precision of observability figure corresponding to control/PVS.

Use can also optionally increase the precision of observability figure for the back-projection approaches of high-order refinement previously discussed.Partly, can also measure determine for the control towards this adaptive refinement of accurate secondary event face by ESO.

By representing to reduce storage demand with middle increment PVS, wherein during precomputation, by identifying corresponding blocking or exposing the district and come significantly to block those important silhouette edges that expose and identify producing of linking up.

The present invention includes polygon or the polygon segmentation specifically looking the unit transition period and expose or be blocked of model of cognition directly.Tabulation for the recently visible polygon of looking the unit transition or polygon segmentation is called as increment G+ sub-grid.Tabulation for the polygon that recently is blocked of looking the unit transition or polygon segmentation is called as increment G-sub-grid.

The present invention includes following methods: look the observability figure of unit and traversal for looking the unit to another as a result observability figure that blocks/expose the border that looks the unit from one by calculating for each, identify for the district and recently expose the coherent district in district of recently being blocked of looking cell translation.The method is used for exposing/blocked area of identification connection.Effectively blocking of these districts is to use the method identical with previously described effective static barriers to measure.In the situation that effectively block to be called as effectively dynamically and block (EDO) in these increments G district.EDO is used for the district of the coherent effective difference observability of identification.

Can according to existing PVS and store come to be paired in observability PVS data of looking the unit next life in common mode for the increment G+ data of looking the unit transition and increment G-data.

Alternately, in the off-line pre-computation phase profile contour that forms so high EDO district is identified and mark.These mark profile contours are bases of the intermediate representation of increment PVS, compare with increment G-sub-grid data with directly storing for each all increment G+ that look the unit transition, and it has reduced storage demand fully.

In this centre observability figure/PVS represents, the profile contour that dynamically exposes or the profile contour that dynamically blocks (the encirclement district of high EDO) are carried out mark.Use comprises for two simple traversal of looking the same observability figure that blocks the border of unit of concrete transition to be identified the high EDO district that links up in the off-line pre-service.

Silhouette markup and triangular mesh data are stored together together with the crossing hint in border of blocking of looking the unit transition for each.Block the border and be the border of the observability figure that looks from the district that is produced by the umbra event face that is incident on the mark profile contour.Mark profile contour and the corresponding border of blocking form polyline.In working time, can construct complete profile contour according to some mark silhouette edges (with the boundary segmentation that blocks of correspondence) with seeking the simple algorithm that connects silhouette edge and polygon-polygon cross-shaped portion.

According to some embodiments, for simple profile contour, often can store this contour by the single beginning limit of only marking whole mark profile contour.Can promptly identify the remaining connection single order silhouette edge that forms this contour in working time.This scheme is so that the intermediate representation of usage flag profile contour is very effective in storage.

According to some embodiments, for compound profile contour (forming when intersecting when being incident on umbra event face and another a profile contour on the simple profile contour), except the compound-contoured summit (CSV) that also will store the cross-shaped portion that represents simple contour, storage scheme is similar.

Use intermediate representation, can look according to father's (comprising property) of first front construction the unit for the observability figure that looks the unit/PVS and generate with the traversal of having simplified.This traversal of father's observability figure marches to the profile contour that blocks of mark, and it supports the new umbra event face that blocks the border that produces, and the above-mentioned new border of blocking gets around the grid elements that recently is blocked effectively.The method needs below eliminating: storage is looked the increment G-information of looking the unit transition and increment G+ information between the unit for the son with shared father.Using significantly, the profile contour that blocks of mark is particularly useful in distributed clients-server implementation in the method for increment ground generation working time observability figure/PVS, this is because client can use the method to remove for the geometric figure that recently is blocked of looking the unit transition, and does not need to receive clear and definite increment G-information from server.

Except being used for directly generating for the observability figure that looks the unit/PVS, mark profile contour can also be used for generating increment PVS data when needed.Use the method, according to some embodiments, (for example do not store increment PVS data, increment G+ and increment G-sub-grid data), but the when needed assurance of usage flag profile contour information, existing observability figure and (for increment G+ sub-grid data) current observability figure comprises for the recently visible geometric superset of looking cell translation and generates increment PVS data.In some embodiments, the superset information of back may be provided in store for comprising that for it specific father-son occuring looks the son of unit transition and look the increment G+ sub-grid data that the father of unit looks the unit.

Use this three data set, be paired in the specific father who looks the unit transition-sub-increment G+ and increment G-data next life by the simplification traversal with the father looks observability figure corresponding to unit to first front construction.Mark profile contour (be associated block border hint) be used for looking the observability figure of unit/PVS according to the father and come constructor rapidly to look the observability figure of unit/PVS.Alternately, can generate clearly increment G+ data and increment G-data by respectively the district of recently exposing and the district that recently is blocked being traveled through.Rear a kind of method is useful in client-server is implemented, and wherein server is to use looking ahead increment G+ and/or the increment G-sub-grid data transmission visibility events server to client based on navigation.

Alternately, only store clearly for the increment G+ data of looking the unit transition and by the simplification of the father being looked the unit and travel through to generate increment G-data.In this is implemented, simplification (fast) traversal of father's observability figure is marched to the profile contour that blocks of mark, it supports the new umbra event face that blocks the border that produces, and the grid elements that recently is blocked is walked around on the above-mentioned new border of blocking effectively.

This increment PVS method representation is used for the effective table demoder that the stream based on observability of external memory geometric figure and texture information is processed, and wherein in the coding of off-line, precomputation the profile contour that dynamically blocks or expose (for looking unit-look unit transition) is identified and mark; And the contour of result's mark and other hint information one are used from according to existing PVS/ observability figure and promptly construct PVS/ observability figure (or increment G sub-grid data) in working time.This codec makes it possible to carry out distributed clients-server implementation, and wherein the storage/transmission cost can optionally reduce to increase to assess the cost working time as cost.

In addition, the coding strategy based on consciousness is used in time not being passed to client to generate for during the complete PVS of forward sight unit/viewpoint low level of detail (LOD) geometric figure and texture information being encoded when increment G+ sub-grid information.This strategy utilizes following true: the human visual system can not differentiate fully with approx less than 1000 milliseconds of information that present.The method is so that make it possible to occur the relatively harmless mis-behave of consciousness during the visible continuity of low space-time: the situation of challenging in a similar fashion codec and human visual system's performance.

The details that discloses in this manual this codec with and adopting the client-server method of looking ahead stream contents processing based on the navigation purposes in transmitting.

Table I c has summed up and has adopted the many art methods that PVS determines and shade calculates that are characterized in the various visibility events faces among the Table I a.Last column of this table comprises the definite current method of increment PVS of looking the unit from looking, and it uses the method that comprises single order and high-order Conservative Linear umbra event face (CLUES) of the present invention.

Table I c

PVS and shadow method

Principle of operation: the model that the single order observability is propagated

According to some embodiments, the observability face of looking from the district has been used in the observability precomputation of looking from the district of this method, and its conservative model of simplification of propagating with the observability that is called as the single order observability is constructed.

Accurate observability in the polyhedron environment is that the secondary visibility events face that is caused by the interaction of the observability between the tlv triple on limit is arranged.By contrast, first order modeling consider by the limit between the visibility events face that causes alternately of observability.Use method disclosed herein, single order observability model produces continuous conservative umbra event face, its can be used for structure conservative from look the observability figure that looks the unit with relevant from looking potential the visible collection (PVS) of looking the unit.

The first order modeling that observability is propagated is based on the simplification to conservative hypothesis: if as seen silhouette edge looks from looking the unit, then as seen silhouette edge from looking all parts of looking the unit.This hypothesis causes the single order silhouette edge is defined as those limits: assembly triangle has a back side pair with respect to the institute of looking the unit, and another assembly triangle is just being faced with respect at least one point of looking the unit, and the assembly triangle does not face with each other.This definition is in fact identical with the definition of the silhouette edge of looking from point, and reflects the following fact: first order modeling will be looked the unit in some importance and be considered as viewpoint.

In the single order observability, any segmentation of supposing silhouette edge is looked and fully is blocked or from looking unit look fully visible (from looking all parts of looking the unit as seen) from looking the unit.That is, in the single order observability, if from the arbitrary portion looking the unit and look as seen silhouette edge, supposes then that as seen it from looking all parts of looking the unit.

First order modeling does not have to explain the effect of blocking along the variation of silhouette edge segmentation, and it is to be caused by silhouette edge and the limit insertion of looking generation secondary three limits (or EEE) the visibility events face between the unit that above-mentioned variation is blocked.Alternatively, single order observability model produces plane visibility events face it is corresponding or conservatively be positioned at the umbra border that normally quadric accurate secondary is looked from the district corresponding to the umbra event face of looking from the district accurate, the plane.The first order modeling that observability is propagated has only adopted the plane visibility events face that is interacted and caused by observability between the pairing on limit.In addition, very near accurate event face, it can be quadratic form for single order, plane visibility events face usually, and in many cases, single order event face is observability (umbra) border of looking from the district accurately.

According to some embodiments, generate single order visibility events face with simple rotation and swept algorithm.In one embodiment, it is protruding supposing to look the unit.This hypothesis has been simplified rotation and has been scanned building method.The rotation and scan method alternate embodiments can according to the structure from the non-protruding single order visibility events face of looking the unit of looking.For example can any non-protruding unit decomposition of looking be become convex assembly by tetrahedralization.

In some embodiments, identify the single order grid silhouette edge that causes single order visibility events face with three standards.In some embodiments, the single order silhouette edge is defined as flowing those limits of passing through following test of shape triangular mesh:

1) triangle sharing this limit is with respect to all back sides, summit of looking the unit pair,

2) another triangle of sharing this limit is just faced with respect at least one summit of looking the unit,

3) the assembly triangle of sharing this limit is the back side pair relative to each other.

Single order Conservative Linear umbra event face (CLUES) is also referred to as wedge, and it has two types.In some embodiments, also conceptually will look the unit and be considered as " source " or light source.

According to some embodiments, one type wedge is that the single order silhouette edge by the summit of looking the unit and grid forms (SV-ME).The wedge of another kind of type is that the inside corner summit by the limit of looking the unit and grid forms (SE-MV).This SV-ME type at first is discussed.

According to some embodiments, in order to construct the SV-ME wedge, single order silhouette edge and the support triangle shape of looking between the unit are identified.This triangle is formed on silhouette edge and looks being called as of unit and support and to look between the certain vertex of summit, unit (SVV).Back side triangle by the test limit be formed on each and look angle between the triangle between summit, unit and the silhouette edge and identify with support corresponding to single order silhouette edge and look the summit, unit.Produce forming with the summit of the leg-of-mutton minimum angles in the back side (that is, maximum negative cosine value)-Bian vertex of a triangle is the first summit that runs in " rotation " on leg-of-mutton plane by silhouette edge, this back side.This looks the summit, unit is to look the summit, unit for the support of the grid silhouette edge of correspondence.

The single order wedge of incident is to be to pass the summit on this limit and two other limits of the line that summit, unit (SVV) is looked in the support corresponding with this silhouette edge form by this limit self and each on single order grid silhouette edge.The profile summit is passed along partly boundlessly extending away from the direction of looking the source, unit from SVV in these two limits.Can say that this wedge is regarded as is formed on the extension that the support triangle shape between the summit, unit (SVV) is looked in silhouette edge and corresponding support.As noted earlier, because the wedge of the type is to form according to the silhouette edge of grid and the summit of looking the unit, so be referred to as source summit-Grid Edge (SV-ME) wedge.

Degenerate case may occur, and the rotation of wherein looking the unit from grid profile by-pass runs into two or more supports that produce the identical anglec of rotation and looks summit, unit (SVV).This occurs in when the limit that comprises SVV of looking the unit is parallel with the grid silhouette edge.In this case, grid silhouette edge and look support triangle shape between the unit and be actually and support quadrangle.This method is processed this degenerate case by constructing special SE-ME wedge.

In some embodiments, rotary manipulation produces the SV-ME wedge for each grid single order silhouette edge.But the visibility events face that is in the place, shared summit of two single order silhouette edges is that necessarily complete cross-shaped portion by two adjacent S V-ME wedges limits.Although adjacent SV-ME wedge always intersects at the place, profile summit of sharing, at the inside corner place of profile contour, these wedges only can intersected by their two a single point places that support silhouette edge to share.In this case, their cross-shaped portion does not form the continuous umbra face of a part of passing the profile contour.The structure of the visibility events face of the profile contour at the place, profile summit that leap is shared depends on how adjacent SV-ME wedge intersects.

According to some embodiments, conceptual reverse rotation operation can be used for determining whether adjacent SV-ME wedge intersects to form continuous umbra face.The operation of oppositely scanning of to the limit line segment that is anchored on SVV being scanned from the summit along the grid silhouette edge of correspondence generates the support triangle shape identical with the support triangle shape that forms in aforementioned rotary manipulation.Conceptive, but can be with the uncontinuity of reverse rotation operation for the visibility events face of identifying the place, shared summit that may appear at adjacent silhouette edge.

If two adjacent grid silhouette edges form " external corners " or the protruding turnings of grid stream shapes, oppositely scan operation and can not run into any restriction of scanning (that is, blocking) to the place, summit of sharing then like this.Thereby the SV-ME wedge corresponding with adjacent " external corners " silhouette edge can intersect to form the continuous visibility events face of crossing over two silhouette edges.The SV-ME wedge that is incident on the adjacent external corners single order silhouette edge can intersect to form continuous like this visibility events face, even rotate to looking different on the unit SVV for the support triangle shape of adjacent silhouette edge.

On the contrary, if two adjacent grid silhouette edges form " inside corner " or the non-protruding turning of grid stream shape, the SV-ME wedge that then is incident on these two limits can not intersect at the place, profile summit of being shared, so that form the continuous visibility events face of crossing over adjacent grid silhouette edge.The support polygon corresponding from adjacent " inside corner " silhouette edge can rotate to looking different on the unit SVV place.In this case, adjacent SV-ME limit still can be intersected at the place, profile summit of being shared, but their cross-shaped portion can not form the continuous visibility events face of crossing over adjacent silhouette edge.Be anchored on oppositely scanning operation and can running into restriction (blocking) at such place, inside corner summit by scanning of silhouette edge of SVV place.This restriction causes the uncontinuity of the visibility events face that formed by adjacent inside corner SV-ME limit.

Continuous visibility events face at such inside corner place can be constructed by the reverse previously described operation of oppositely scanning at the inside corner place.This scans and is anchored at now shared place, inside corner grid profile summit, and scan along the silhouette edge of looking the unit, these limits are with respect to the silhouette edge of inside corner grid profile summit for looking from point, scan to start from for the SVV of one of grid silhouette edge and end at SVV for adjacent grid silhouette edge.Each scans looks unit silhouette edge and inside corner summit and forms and scan triangle.This leg-of-mutton limit of passing corresponding grid outline polygon extension defines wedge.Because being the summit by the limit of looking the unit and grid, such wedge forms, so they are called as the SE-MV wedge.Scan the set that operation can produce the SE-MV wedge along (looking from point) profile contour of looking the unit such, it has formed the continuous visibility events face of the SV-ME wedge of connection (or disconnection) adjacent mesh silhouette edge.

Conceptive, then conceptual when oppositely scanning the inside corner place that operates in single order profile contour and running into restriction (blocking) when what be anchored on the SVV place, oppositely scan operation and be reversed.This has oppositely produced the actual operation of scanning, and it has constructed the SE-MV wedge that scans triangle and correspondence that forms continuous visibility events face (single order umbra event face), and it connects the SV-ME wedge from adjacent single order grid silhouette edge.This scans operation and generates the SE-MV wedge, above-mentioned SE-MV wedge is as on the summit that is located at grid profile contour and reflected the visibility events border, and this border is mainly by determining the combination of " blocking " of being reflected in the silhouette edge place in the SV-ME wedge and comprising in the SE-MV wedge that is incident on the profile summit that be reflected in of looking the viewpoint on the unit.

Should be noted that for the protruding unit of looking, two paths looking the unit silhouette edge of connection generally can be connected to another SVV with a SVV.Only the SE-MV wedge chain that has for a moment scanned connection adjacent S V-ME wedge in these paths has the continuous visibility events face of consistent face orientation with formation.In some embodiments, this concrete path is called as to support and looks unit profile contour (SVSC).Other places have in this manual proposed the test of identification SVSC.

According to some embodiments, for the structure of single order wedge, can test to replace to detect with another restriction of the observability at inside corner grid vertex place conceptual oppositely scanned operation.This test comprises the normal of comparison adjacent mesh silhouette edge.Be oriented to so that the normal that they are relative to each other just being faced if the grid silhouette edge of two connections has, then the summit of sharing be called the external corners of grid.

According to some embodiments, when running into inside corner grid profile summit, then generate the single order wedge that passes this summit by scanning operation, wherein be anchored on scanning of inside corner grid profile summit place and be along support and look unit profile contour (SVSC) from the SVV corresponding with silhouette edge to carrying out with SVV corresponding to another silhouette edge, thereby generate the SE-MV wedge.

The operation of scanning that generates the SE-MV wedge is only carried out at the inside corner place of profile contour.Carry out this at place, corner profile summit, the outside and scan the crossing redundant SE-MV wedge of adjacent S V-ME wedge that can generate and only be in this profile summit, so they do not facilitate the continuous umbra event face of the profile contour that supports.

As previously mentioned, SE-MV wedge " inside corner " that can appear at the profile contour that is called as simple contour that the connection silhouette edge by single grid forms located.More generally, the SE-MV wedge can be incident on any " inside " or the non-concave edge of polyhedron polymerization umbra (PAU) face.These " inside corner " features can also be formed on when the wedge from two profile contours (belonging to identical grid or different grids) intersects.Cross-shaped portion from the wedge of a contour and non-adjacent silhouette edge is called as combination or compound-contoured summit (CSV).In the research of smooth manifold, such cross-shaped portion is called the t joint.At t joint cross-shaped portion place, the wedge of a silhouette edge and non-adjacent silhouette edge (from identical or different contour) intersect.This t joint cross-shaped portion generally occurs in the following manner: the SV-ME wedge that intersects mutually disjoints to form continuous event face in the t joint.The cross-shaped portion of the result's of two the SV-ME wedges in single order silhouette edge place degeneration represents CSV.

In the existing method of such CSV, adopt with aforementioned scan operation identical, be anchored on scanning of CSV now and operate and generate the set that the SV-ME wedge of other disconnections is connected into the SE-MV wedge of continuous conservative umbra event face.Discuss in detail such as the another part at this instructions, usually, accurate umbra event face is high terrace (for example, quadric surface).Present invention resides in and carry out on the CSV aforementionedly scanning operation so that the wedge of the constructing approximate high terrace that is incident on the reality on the CSV conservatively.

Theory of operation: the relation between single order visibility events face and the support package

According to some embodiments, the first order modeling that observability is propagated has adopted the new geometric figure structure that is called as support package.

According to some embodiments, the support package between the stereoscopic unit of multiaspect and the polyhedron grid object is polyhedron volume, and it comprises all possible sight line of looking between unit and the mesh object.This support package is the tetrahedron that is surrounded by the support polygon of looking between unit and the mesh object.If all be protruding depending on unit and mesh object, then support package identical with convex closure and its can use familiar gift wrap algorithm (O ' Rourke, Computational Geometry in C Second edition Cambridge University Press1998) to construct.In some embodiments, if looking the unit is but that protruding mesh object needs not to be protruding, then can form support polygon with following algorithm.

The silhouette edge that each single order of mesh object is looked from the district is identified as following those limits: have one with respect to the right assembly triangle in all back sides, summit of looking the unit and another assembly triangle of just facing with respect at least one summit of looking the unit, and the said modules triangle back side pair relative to each other.For in these single order silhouette edges each, the support polygon that is incident on this limit sound is constructed on the summit that rotates to the minimum anglec of rotation of formation of looking the unit by the leg-of-mutton normal direction of back side assembly from this edge.Be called as and support this summit look summit, unit or SVV and form the support polygon (being generally triangle) that is incident on the silhouette edge with the end points of single order silhouette edge.The support polygon of the type is called as SV-ME(source summit-Grid Edge) support polygon.

If this looks the summit, unit is the end points of looking element sides parallel with the mesh object silhouette edge by chance, then should rotation can runs into two of forming equal angular and look the summit, unit.In this case, support polygon is the quadrilateral (that is, SE-ME support polygon) that forms by looking element sides and mesh object silhouette edge.By forming the limit that comprises mesh object and all support polygons of looking the summit of unit to supporting the rotation of looking unity element.

If adjacent mesh object silhouette edge produces rotation to the identical support polygon of looking the summit, unit, then support polygon is being intersected with the common edge place that the mesh object profile summit that should share forms by this summit.In this case, the support package at place, mesh object profile summit is limited by two support polygons of person fully.Adjacent mesh object outline limit can also produce rotation to the support polygon on the different summits of looking the unit.In this case, two support polygons do not form continuous face at place, grid profile summit.For the support envelope near place, this summit, between grid profile summit and the certain edges thereof of looking the unit, construct one or more support polygons." scan operation and carry out this structure: scan along each chain of looking the unit silhouette edge looked between the profile summit, unit that rotates the adjacent mesh silhouette edge by aforesaid.During this scans, form support polygon according to these each limit and grid profile summits of looking in the unit silhouette edge.This structure can be regarded as " the scanning " of looking unit profile side chain and scan polygon so that look unit silhouette edge generation for each.Usually, look scanning between the summit, unit for two and can take a more than paths, but only a paths can scan one group of polygon of two original support polygons of connection has consistent face orientation with formation continuous face.This path is to support to look unit profile contour (SVSC).

This algorithm has produced to use and has surrounded or protection support lattice object and the continuous polygon facet of looking the unit.In some embodiments, if look unit and mesh object both for protruding, then only intersected at their place, limit by the support polygon by this algorithm construction and form the convex closure of looking unit and mesh object.

If be non-protruding depending on the unit, then from the inside corner summit of stream shape grid single order profile, can be the complicated contour that comprises most advanced and sophisticated and t summit depending on the profile contour of looking from point of unit.If mesh object also right and wrong is protruding, then support polygon can intersect in their inside.

But, being restricted to convex polyhedron if look the unit, the profile contour (observing from inside corner grid profile summit) of looking from point of then looking the unit often is the simple contour with tip or t summit.Thereby, be protruding if look the unit, then simplify fully looking the operation of scanning of unit contour.According to some embodiments, be restricted to convex polyhedron and simplify fully and scan operation by looking the unit.

Be incident on single order wedge on the single order grid silhouette edge and be formed in the extension that the corresponding support polygon between the summit, unit (SVV) is looked in same grid silhouette edge and support.The wedge of the type is that the line by two extensions of intersecting in the grid silhouette edge of grid silhouette edge (that is, line segment) and support polygon consists of.Thereby this wedge as initially constructing, is easy to extend until intersect with the grid polygon away from looking semiinfinite ground, unit.The wedge of the type is extended to form and is called as the SV-ME wedge by the SV-ME support polygon.

The single order wedge that is incident on single order mesh object profile inside corner summit is the extension of scanning triangle (that is, be formed on grid profile summit and look SE-MV support polygon between the limit of unit profile contour).The wedge of the type is to be constructed by two lines that intersect in this summit of grid profile summit and support polygon.These two lines extend to form the border of SE-MV wedge away from looking semiinfinite ground, unit.Thereby this wedge is easy to extend until intersect with the grid polygon away from looking semiinfinite ground, unit.Because the wedge of the type is to be formed by (that is, depending on the unit) limit, source and grid vertex, so be referred to as the SE-MV wedge.

The SE-MV support polygon that is incident on the external corners summit of grid profile contour is the encirclement polygon of looking the reality of the support package between unit and the grid profile protruding.But the extension of this support polygon can produce the SE-MV wedge that only intersects with single order umbra event face tangently on corner profile summit, the outside.Thereby such wedge can not facilitated single order umbra event face/volume and not need it is constructed.

Following special circumstances occur: the rotation of single order grid silhouette edge is to (that is, being supported by the SVV) SVV for the summit of looking element sides parallel with the grid silhouette edge.In this case, grid silhouette edge and the support polygon looked between the element sides are quadrilaterals.Such support polygon and its corresponding umbra event wedge are called as SE-ME support polygon, wedge.Identification SE-ME wedge is useful, and this is because be different from the wedge of other types, and seeking itself is the visibility problem of (looking from segmentation) of looking from the district for the SE-ME wedge about the visible cross-shaped portion of wedge.Compare for SV-ME and SE-ME wedge look from point, about the observability solution of wedge, SE-ME is more complicated a little about the observability solution of wedge.

That every contour is the simple contour of polyline to mesh object and the single order profile contour that supports of description hypothetical trellis object of looking the front of the support package between the unit.In fact, in fact the profile that any single order is looked from the district can be compound contour, and wherein whole contour is to form by intersecting contour.Contour is intersecting from the crossing place, place of the wedge of a contour and another contour (that is, single order silhouette edge).Should crossing appear at compound-contoured summit (CSV) locates.Interact if consider the high-order of limit tlv triple, then these CSV are general corresponding with quadric surface.Based on the rotation of single order observability model and this method of scanning structure CSV is considered as simple inside corner profile summit; At the one or more SE-MV of each CSV structure, create the conservatively approximate accurate quadric continuous polygon umbra event face that is supported by the single order silhouette edge.

By using SV-ME(and SE-ME in particular cases) and SE-MV support polygon/umbra wedge, compare the method for the non-penumbra of linearization that the convex closure and excessively having estimated significantly on the calculating SVE-ME plane of Teller blocks, the embodiment that comprises this method provide to reality from looking the more accurate myopia of the umbra event face of looking the unit.

Be different from non-half image method of linearization, this rotation and the method for scanning are not limited to the more limited visibility problem that passes the entrance sequence.

In some embodiments, in order to construct the umbra uncontinuity grid of looking from the district or the observability figure that looks from the district, determine the polygonal visible cross-shaped portion of single order wedge and grid.The identification of the visible cross-shaped portion of wedge and mesh triangles shape is called as the visibility problem of " about wedge ".Embodiment comprises that 2D(namely, on wedge) 1-stream shape (polyline) traversal method, wherein the structure of visibility events line (that is, the event of 1 degree of freedom surface) with 1-stream shape traversal and disturb check staggered to produce for the output sensitivity solution about the observability of wedge.

The method that this stream shape traversal method is expanded to the traversal 2-stream shape (that is, triangular mesh) among the 3D comprises from looking the look observability figure that looks from looking the unit of visible grid polygon segmentation of unit with structure.From observability figure, derive PVS.The aforesaid 2D(1-stream of this 3D trellis traversal method call shape) trellis traversal is processed the observability that solves about wedge.

Suppose under the first order modeling that observability is propagated, be called as single order polyhedron umbra volume from looking the unit spatial volume that is blocked by mesh object of looking.Because may intersecting with polymerization, each umbra volume blocks, so these volumes are called as single order polyhedron polymerization umbra (PAU).

The single order PAU that is also referred to as PAU is surrounded by the polygon that is called umbra boundary polygon or UBP.These polygons form by single order wedge and triangular mesh polygon and with the cross-shaped portion of other single order wedges.PAU is also surrounded by the visible grid polygon of single order segmentation (that is, comprising from looking the segmentation of the observability figure that looks the unit).Although UBP and the segmentation of visible grid polygon form the continuous umbra face closed, that limit the border of PAU that needs not to be together.

As described in detail in conjunction with 3D2-stream shape traversal method (Figure 20 and relevant figure), according to some embodiments, the structure of observability figure may further comprise the steps: determine to be actually the no PAU of being positioned at volume about the visible ambiguity line segment of wedge, also therefore look and be blocked from the whole unit of looking.The method comprises and can answer questions next time in polyhedral some pardon test of the modification of this inquiry of single order PAU in the situation of constructing ambiguously PAU.

In the observability method about wedge, use in 2D, use 1-stream shape polyline traversal method (Figure 15 and relevant figure) be for structure from look the observability figure that looks the unit, in the better simply enforcement of the 2-of 3D stream shape traversal method.

Theory of operation: expression is from looking the method for the observability of looking the unit

Embodiment adapts to from looking three kinds of different expressions of the observability of looking the unit.In Table II, show the feature of these three kinds of expressions and its art methods with the observability that represents with uncontinuity grid completely to look from the district is compared.

From a kind of expression of the observability of looking the unit and looking, use polyhedron polymerization umbra (PAU) at Conservative Linear, to identifying from looking the unit spatial volume that is blocked of looking of reality.These volumes are by according to surrounding from looking the umbra boundary polygon (UBP) that umbra wedge that unity element looks becomes.In fact above-mentioned wedge intersects with the grid polygon and intersects each other to determine UBP.Although most of shade volume methods are the shades of looking from point, should represent suitable with the umbra volume product representation.

At Conservative Linear from the another kind of the observability of looking the unit and looking represents, the conservative discontinuous grid of line linearization umbra (CLUDM), mutually disjoint from looking the umbra wedge that unity element looks, and only intersect to form conservative uncontinuity grid with the grid polygon, wherein the district of grid is corresponding to fully visible district, this shadow zone or non-this shadow zone.Non-this shadow zone is actually looks the unit from its part of looking visible penumbra region one type.Additional test is used for distinguishing this shadow zone and non-this shadow zone (for example, definite PVS that looks from looking the unit).

From the third expression of the observability of looking the unit and looking, according to some embodiments, in Conservative Linear umbra uncontinuity observability figure (CLUVM), only express complete visible district and this shadow zone at Conservative Linear.Because in this case, PVS is corresponding with complete visible district, so this is the expression that is particularly useful.By determining by potential the blocking the border and whether construct CLUVM from looking the umbra border of looking the unit of each of the umbra wedge one-tenth of look from looking unity element (that is, or limit).Other places in instructions show this details of determining and the output sensitive method of constructing CLUVM.

Compare from these three kinds of expressions of looking the observability of looking unit art methods with completely uncontinuity gridding conservative.In uncontinuity grid completely, most of borders comprise the penumbra region, and it is to look the unit from its part of looking visible district.Usually, the part of unit is not looked from this shadow zone of its reality of looking visible by the district of much smaller quantity.The penumbra region of uncontinuity grid and this shadow zone can be surrounded by line segment and/or quafric curve completely.Such as what in the art methods of incomplete uncontinuity gridding, propose, only use linear modules to cause discontinuous umbra border and the observability that therefore can not be used for determining looking from the district.

For disclosed a large amount of reasons in other places in this manual, the accurate event face that Conservative Linear umbra event face (CLUES) adopts in the art methods far fewer than completely uncontinuity gridding.Thereby the approximate complexity of the step of CLUDM is far below the complexity of uncontinuity grid completely.In fact, use the responsive building method of output of the present invention, complexity (constructing and store both) generally only determined by the quantity on visual profile limit, as for the CLUVM in the Table II by N _v ⁴Represented.

In Table II, provided the estimation of these complexities, and other places are in this manual discussed to it in detail.

Table II

The table that the expression Conservative Linear is compared from the method for the observability of looking the unit and looking and classical uncontinuity gridding

^*The quantity of supposing the single order silhouette edge is the quantity on O(limit) ^1/2

Theory of operation: the high-order Conservative Linear umbra event face of single order back projection

According to some embodiments, single order observability model hypothesis: for any support polygon of looking between unit and the single order stream shape grid profile, the Bian Congyu corresponding with the single order silhouette edge of support polygon supports the summit of looking support polygon corresponding to summit, unit (SVV) look fully as seen (being blocked).That is, for the SV-ME wedge, suppose that corresponding support triangle shape is not crossing with any other the polygon when the arbitrary portion that can block corresponding grid silhouette edge when corresponding SVV observes.Similarly, for the SE-MV wedge, suppose corresponding scan triangle not with any other polygon that when from the observation of corresponding inside corner grid single order profile summit, can block the corresponding arbitrary portion of looking contour limit, summit, unit.

In fact, the support polygon corresponding with wedge can fully be blocked, fully is not blocked or partly is blocked.If support polygon is not blocked fully, then corresponding single order wedge is by the grid wedge or pushes up point-supported accurate visibility events border.If support polygon is blocked fully, then whole wedge keeps the conservative of this border is similar to.If support polygon partly originally blocks, then the part corresponding with the segmentation that is not blocked of support polygon of wedge is accurate visibility events border, and the part corresponding with the segmentation that is blocked of support polygon of wedge is that the conservative of exact boundary is similar to.

Following part has been summed up following methods: coming adaptively with back projection, refinement single order wedge interacts to explain the high-order observability that exists when support polygon completely or partially is blocked.Back projection is the processing of determining from the part in specifiable lattice element (that is, the single order silhouette edge) source of looking visible (that is, depending on the unit).According to some embodiments, in order to calculate back projection, adopt the single order observability models and methods of silhouette edge as linear light source.

Described method has adopted the first order modeling of the simplification that causes linearization visibility events face of observability propagation so far.These First-Order Mode profiles are surrounded by the single order wedge that generates by the method for rotating and scan.

These single order wedges have two types: SV-ME wedge and SE-MV wedge.Reflected by the limit visibility that the combination of blocking with the silhouette edge of grid causes that comprises corresponding to the viewpoint of looking the point on the unit by the SV-ME wedge that generates to the rotation of looking the summit, unit from Grid Edge.The SE-MV wedge that scans generation that runs through the limit of looking the unit by the point from grid has reflected by the caused limit visibility that comprises on the limit of looking the unit (that is, border).Under single order observability model, SV-ME(namely, SE-ME in particular cases) and the SE-MV wedge only be the visibility events face that occurs in the polyhedron environment.

This single order wedge of two types can by to grid with look corresponding between the unit support polygon and extend to construct.The important hypothesis of single order observability model is that any single order grid silhouette edge is to look and fully as seen or fully be blocked from looking the unit.This is identical with the following stated: for any single order silhouette edge, suppose that looking the unit looks from the limit and fully be blocked or fully as seen.

Similarly, first order modeling hypothesis: when looking the summit, unit from the support of correspondence or the limit is observed, the silhouette edge that supports or summit are blocked fully or are not blocked fully.

According to some embodiments, use the single order rotation and scan method, for example, if the segmentation of single order silhouette edge is not blocked, suppose that then segmentation blocked (that is, less than crossing with any other grid polygon) fully with support triangle shape between the corresponding SVV.In fact, if this support triangle shape is not blocked fully, then first order modeling be accurate and corresponding SV-ME wedge be by the grid silhouette edge support from looking accurate the assembly of the umbra event boundaries of looking the unit.But if this support triangle shape is blocked partially or completely, then first order modeling is that the actual visibility events face that is similar to and is incident on the silhouette edge can be comprised of quadric surface and the plane of intersecting.In addition, single order silhouette edge (or its segmentation) even may not support actual visibility events face.Alternatively, in fact actual visibility events face can be produced by other limits that are called as the high-order silhouette edge, so that in fact all or part of single order silhouette edge is positioned at the inside on visibility events (that is, umbra) border and also therefore is blocked.

Embodiment comprises following methods: scan to identify the element that blocks that causes high-order visibility events face by what carry out corresponding support triangle shape, identifying the single order hypothesis is coarse silhouette edge and summit for it.By calculate to look the unit in fact from the back projection that silhouette edge is looked or the visible part in profile summit is identified, be similar to high-order visibility events face.This back projection self is the visibility problem of looking from the district that uses single order rotation or the method for scanning to solve.Use the method, the conservative single order wedge of adaptively refinement is similar to corresponding accurate high-order visibility events face to desired error margin.

In some embodiments, higher-order method is implemented as following technology: the such face of the accuracy of test single order visibility events face and modification or " adjustings " is more accurately to be similar to relevant high-order observability face.Single order visibility events face is incident on the silhouette edge that single order looks from the district.The single order silhouette edge defines the conservative profile contour of grid.Accurate high-order observability umbra event face not necessarily is incident on the single order silhouette edge and can appears on other Grid Edges that are called as the high-order silhouette edge.Compare the event face of the correspondence that is incident on the single order silhouette edge, the high-order visibility events face on these high-order silhouette edges of being incident on can produce significantly and more block.In fact, usually, in fact the event face that occurs from the high-order silhouette edge can surround the volume that blocks that comprises corresponding single order silhouette edge.

Embodiment comprises following methods: so that the event face of regulating keeps incident from the single order silhouette edge, be similar to high-order visibility events face by " adjusting " single order visibility events face.Part has subsequently been introduced following methods: identify when the single order silhouette edge has reduced the precision of the high-order event face that calculates significantly when high-order visibility events face is restricted to.Other embodiment comprises the method for the specific high-order silhouette edge of identification support visibility events face, and it is the visibility events face of approximate exact more accurately.

The below is about the description that occurs high-order visibility events face where and how on polyhedron grid object.This framework provides and has used face of a polyhedron self-adaptation, the basis of the completely new approach of approximate these high terraces progressively.

For concept is described, substitute the better simply situation of area source as beginning with line source.Envision single line source and comprise single segmentation and single convex polyhedron.Because polyhedron is protruding, self do not block or inside corner so do not exist.Thereby polyhedral umbra uses aforesaid single order rotation and swept algorithm accurately to form.In this case, each single order silhouette edge of grid supports by the summit, support source (SVV) to the correspondence in source and forms single SV-ME wedge, and it is line segment in this case.

Now, the imagination is violated single order and is supposed to form this silhouette edge for the specific single order silhouette edge of grid, and the SVV of the correspondence on the line segment of source is sightless (that is, being blocked fully).This occurs in following situation: the support triangle shape and other polygons that are formed by profile Grid Edge and SVV intersect so that do not have expedite sight line between SVV and grid silhouette edge.In this case, because not even not as seen corresponding SVV looks from silhouette edge, so blocking of this axle represents that the single order wedge is not accurate umbra border for the grid silhouette edge.

Can by the distance of identification on the line source for the support on limit look the unit nearest (namely, " rotation to "), but in fact from the grid silhouette edge point of looking visible, obtain better approximate to the visibility events face that is incident on the reality on the grid silhouette edge.For the grid silhouette edge this point is called visible support and looks summit, unit (VSVV).VSVV is positioned at (that is, being positioned at expression looks on the line segment in light unit/source) on the face of looking the unit.It is the visible point of grid silhouette edge that can rotate to from SV-ME UBP.Corresponding SV-ME wedge is the visible event face of umbra that is formed by line source and grid silhouette edge.

Because VSVV provides not too extreme " the watching " of passing the grid silhouette edge and being positioned at grid " back ", compare the single order SV-ME wedge of answering, this high-order SV-ME wedge has produced larger umbra volume significantly.

According to some embodiments, calculate this visible support for the grid silhouette edge and look summit, unit (VSVV) by the grid silhouette edge being considered as line source.In the method, rotate and the method for scanning is used for constructing observability figure about the face of looking the unit with specific grid silhouette edge as light source.In back projection is processed, identify the single order silhouette edge about grid silhouette edge and the polygon of grid between two parties looked between the unit.Construct the single order wedge along the direction of looking the unit at these silhouette edges.These event faces have caused about looking the observability figure of unit, and it will be looked dividing elements and become from grid silhouette edge look visible assembly and sightless assembly.Depending on the visible assembly of unit, to be incident on the point that now can rotate to as the SV-ME wedge on the original mesh silhouette edge of back projection light source be the VSVV corresponding with this grid silhouette edge.

Suppose line source is located so that it seems " over top " at the mesh object at the grid silhouette edge place of discussing.Also suppose in the situation that this is concrete, the observability of looking from the grid silhouette edge of line segment light source is blocked the single leg-of-mutton impact between two parties of support triangle shape (that is, supporting the 2D axle of looking between summit, unit and the grid silhouette edge).In addition, suppose this between two parties the whole four sides axon that more formed by line segment light source and grid silhouette edge of leg-of-mutton single end bay so that triangle " sagging " is in this four sides axon between two parties.In addition, suppose the light source limit, leg-of-mutton limit and grid silhouette edge are mutual obliques between two parties.Each some place on source and silhouette edge, this single limit between two parties affects the mutual observability on two other limit.

The combination expression like this on the limit of these three obliques: the actual visibility events face that is incident on the grid silhouette edge comprises quadric surface.This is classical EEE event TELLER(1992).However, in this case applied back projection rotation and swept algorithm still can be identified the single conservative VSVV on the light source.Rotation from the grid silhouette edge to this VSVV defines and is incident on the conservatively approximate quadric single SV-ME wedge that is incident on this silhouette edge that is incident on the reality on the silhouette edge.In addition, can come more accurately to be similar to high-order (secondary) the visibility events face that is incident on the reality on the grid silhouette edge by segmentation grid silhouette edge and the VSVV that calculates for every sub-segmentation.In this Subdividing Processing, adjacent contour segmentation may produce the different VSVV during the back projection.But corresponding SV-ME wedge is not shared the limit that shares is connected by the SE-MV wedge, above-mentioned SE-MV wedge be by pass from the summit of adjacent contour segmentation line source from a VSVV to another VSVV scan form.In this way, the alternate sequence by SV-ME wedge and SE-MV wedge conservatively is similar to secondary visibility events face.

In some cases, use the grid silhouette edge as the rotation of light source and scan processing and can not produce the single VSVV that looks on the unit.For example, if run into the inside corner of profile contour during back projection, it is single continuous contour form or as CSV, visible " extreme value " feature of then looking the result on the unit may not be point but with as the parallel line segment of the grid silhouette edge of light source.This occurs in following situation: back projection SE-MV wedge be by be anchored on the inside corner place run through grid silhouette edge (as light source) scan generate.This wedge with look the unit and intersect so that cross-shaped portion is support feature (that is, two of the wedge cross-shaped portion end points is VSVV).This situation and front are in simple forward direction single order rotation and to scan the situation of middle description similar, and wherein rotary manipulation causes supporting and looks element sides (SE-ME limit) (for example, the single order silhouette edge is parallel with the extreme value limit of looking the unit).In both cases this high-order forward direction SE-ME wedge structure is similarly managed.

Use is disclosed in the detailed description part of this instructions for the details of the high-order visibility events surface construction that the back projection of the generalized case of polyhedron light source is processed.Usually, rear is identified VSVV with the segmentation on grid silhouette edge or these limits as the single order rotation of line source and the method for scanning.These VSVV usually look unit contour VSVSC by visible support and connect.By scanning of VSVSC processed the SE-MV high-order wedge of constructing between two parties.Other embodiment comprises: structure high-order SE-MV wedge in the situation that disconnects the VSVSC corresponding with adjacent silhouette edge.

According to some embodiments, this back-projection approaches is used for calculating the single high-order SV-ME wedge for grid single order silhouette edge, and it is the approximate very complicated visibility events face that is incident on the cross-shaped portion that can comprise a plurality of quadric surfaces and plane on the grid silhouette edge conservatively.Under these circumstances, can segment the grid silhouette edge, and antithetical phrase segment applications back projection is with the more approximate event face that passes the reality that single limit changes fully.This segmentation can be carried out adaptively based on simple test, and above-mentioned simple test chart shows that linearization event face departs from from the actual visibility events face maximum possible along concrete segmentation.Such as Teller(1992) and people's the method such as Nirenstein(2005): the whole set of at first calculating the event face on the silhouette edge that is incident on, then by with certain type comprise test or high latitude CSG determines which is actual umbra boundary surface, the calculating that the method needs are less.

As before running into for single order observability figure structure, in some cases, for the SV-ME wedge of adjacent silhouette edge or segmentation be disconnect and must be connected by the SE-MV wedge that forms that scans that passes the boundary profile contour of looking the unit by the shared point from the limit so that this scans two VSVV that connect for the grid silhouette edge that connects.

In the single order situation, two SVVs corresponding with adjacent silhouette edge always are positioned on the actual boundary of looking the unit and by the single boundary profile contour of looking the unit and connect.In high-order back projection situation, two VSVV can or can not be positioned on the identical contour.If non-intersect from the adjacent edge two parts of looking visible depending on the unit, then VSVV does not connect by single contour.In this case, the convex closure of two contours can be used for conservatively connecting two high-order wedges and can conservatively generating according to the contour of this connection high-order SE-MV wedge.

According to some embodiments, only when as shown by blocking of the 2D axle between these two structures, corresponding support is looked summit, unit (SVV) and is looked when being blocked from the grid silhouette edge, just back-projection approaches is applied to the grid silhouette edge.Blocking for this of the 2D axle of SV-ME wedge is the visibility problem of looking from point, and it can calculate by the 2D version of previously described trellis traversal algorithm.The EVV of silhouette edge is that visible any segmentation does not need to use backprojection for it, and this is that the single order wedge is accurate visibility events face because for these segmentations.

In addition, according to some embodiments, segmentation and the recursive projection of the contour segmentation that look from it SVV or VSVV are blocked are instructed by following simple test: the current wedge that calculates and the maximum possible deviation that is incident between the visibility events face of the reality in this segmentation are measured.This test is by carrying out to looking the unit along the reverse direction rotation that is generally used for seeking SVV from this contour segmentation.Along the rotation contour identification limit of this direction and look separating plane between the unit.This separating plane is corresponding to the maximum possible degree that is incident on the high-order observability plane in the silhouette edge segmentation.It is also corresponding to segmentation and look extreme value penumbra border between the unit.In some embodiments, only look when being blocked from the contour segmentation of correspondence when the whole unit of looking almost, the high-order block surface just can be similar to this plane.This penumbra plane and currently represent the current conservative event face at this profile place and the maximum possible deviation between the actual event face for the angle between the conservative SV-ME wedge of this segmentation.Intersect on two planes at the silhouette edge place of discussing and be formed on supported wedge on the gamut of length of this segmentation.The volume of this wedge has reflected current conservative volume and the actual maximum possible deviation that is blocked between the volume of being blocked on this silhouette edge.

Should be noted that this deviation is along with reducing apart from looking the distance of unit.This has reflected the following fact: in larger distance, the visibility events face of looking from the district approaches the visibility events face of looking from point.Thereby the impact of high-order observability is being not too important apart from what look the unit than distant location.In some embodiments, depend on observability and this umbra/penumbra measured value of corresponding SVV, adaptively silhouette edge is segmented.Use the method, according to some embodiments, only enlarge significantly when being blocked volume when compare high-order visibility events face with better simply single order event boundaries, just generate high-order visibility events face.

Aforementioned discussion hypothesis back projection is processed the wedge that is used on the single order silhouette edge that refinement is incident on grid.In fact, back projection processed be applied to the single order silhouette edge and can produce following SV-ME wedge: when triangle that VSVV and silhouette edge by correspondence form is positioned on two leg-of-mutton rear side of shared this silhouette edge, then above-mentioned SV-ME wedge has been violated local observability.In some embodiments, such SV-ME wedge remains the conservative expression of the visibility events face that is incident on the reality on the single order grid silhouette edge.But the violation of so local observability represents: corresponding single order grid profile reality is not the silhouette edge of looking from looking the unit.Alternatively, it is positioned to come from actual compares the single order silhouette edge on the side that is blocked of another visibility events face of looking the silhouette edge of looking the unit, this reality nearer apart from looking the unit from looking the silhouette edge of looking the unit.The type is called high-order grid silhouette edge from looking the silhouette edge of looking the unit.

The general silhouette edge of looking from the district can or can not support high-order visibility events face.As because Drettakis(1994) and Nierenstein(2005) defined, the general silhouette edge of looking from the district is any Grid Edge for the silhouette edge of looking from point of looking any point on the unit.Compare the single order silhouette edge, this comprises that generally grid is polygonal more polygon.

The general grid silhouette edge of looking from the district depends on the general silhouette edge of looking from the district and the geometric accurate layout between two parties of looking between the unit can or can not cause the umbra visibility events face of looking the unit from looking.Can be with identifying the general grid silhouette edge of looking from the district with the slightly different criterion of the criterion of identification single order grid silhouette edge.According to some embodiments, if satisfy following three criterions then the limit is the general silhouette edge of looking from looking the unit: 1) it must have at least one assembly triangle of just facing with respect at least one summit of looking the unit, 2) it must have at least one assembly triangle and 3 right with respect at least one back side, summit of looking the unit) the said modules triangle relative to each other must the back side pair.

Can to aforementioned 3D trellis traversal algorithm make amendment to comprise be incident on non-single order, general from looking the umbra event face on the silhouette edge of looking the unit.In one modification, the 3D trellis traversal is initially carried out with usual way: check that each Grid Edge is to determine whether it is the single order silhouette edge.Carry out back projection with single order grid silhouette edge as line source and calculate the high-order wedge that is incident on the single order grid silhouette edge with VVS and the VSVSC that looks by identification on the elemental area.If corresponding high-order SV-ME wedge has been violated local observability, then identify general from looking the profile contour of looking the unit by following processing: come grid is traveled through until run into one or more general silhouette edges of looking from looking the unit away from the single order limit, it comprises that the high-order visibility events face to blocking original single order grid silhouette edge supports the profile contour of (that is, passing through back projection).When also violating local observability, the high-order wedge can repeat this retraction.This modification starts from conservative result and will guard the precision that the result is refined to expectation based on the measurement result of the maximum deviation of current event face and actual event face.

Compare Teller(1992) non-half image method of linearization, linearization back-projection approaches of the present invention provides more approximate to high-order visibility events face.Non-half image method of Teller has used the rotation strategy from the entrance limit to source inlet, and it has identified the VSVV on the source inlet corresponding with target entrance limit effectively.This point and source inlet limit one are used from and limit the plane that surrounds non-penumbra volume.

These planes are corresponding to the plane of the SV-ME wedge/UBP that is defined by present embodiment.Pointed such as this situation to single order observability (for example, two entrance sequence between) of front, Teller only is similar to the observability border with the SV-ME plane, and the present invention use SV-ME polygon and SE-MV polygon (for example, UBP) both.Compare the non-penumbra based on intersecting plane of Teller, present embodiment produces more approximate to the observability border of reality usually to the use of these polygon wedges.In addition, this method has defined the systems approach for the linearization back projection, comprises trellis traversal, silhouette edge identification and self-adapting subdividing, and it can be applied to the general visibility problem of looking from the district.By contrast, non-half image method of Teller depends on the simple rotation strategy that only can use with the more limited visibility problem that passes through the entrance sequence.

Description, wherein similar Reference numeral refers to the identical or counterpart that runs through some figure, and Figure 55 is according to the exemplary unified processor computer system for implementing to determine with Conservative Linear umbra event face the observability of looking from the district and the system and method that carries out processing based on the content flow of increment PVS of the present invention.Computing machine 5500 is implemented the method for embodiment disclosed herein, and wherein counter body 5502 holds the motherboard 5504 that comprises CPU5506 and storer 5508.Computing machine 5500 also comprises a plurality of input equipments, for example, and keyboard 5522 and mouse 5524 and the video card 5510 that is used for control monitor 5520.In addition, computer system 5500 comprises that further floppy disk 5514 and other can move media equipment (for example, CD 5519, tape and the magnet-optical medium of being connected), hard disks 5512 or use for example other fixing high-density medium drivers of connecting of SCSI bus or enhancement mode IDE bus of suitable equipment.Although CD 5519 is shown the CD box, CD 5519 directly can be inserted in the cd-rom player that does not need box.Be also connected to identical device bus or another device bus, computing machine 5500 can additionally comprise disc read/write device 5518 or CD jukebox.In addition, the printer printing tabulation that picture element also is provided, exposes event and 3-D view.

This system further comprises at least one computer-readable media.The example of this computer-readable medium is CD, hard disk, floppy disk, tape, magneto-optic disk, PROM(EPROM, EEPROM, flash memory PROM), DRAM, SRAM etc.The present invention includes and be stored in the hardware that is used for control computing machine 5500 in any computer-readable mechanism or its combination and be used for so that computing machine 5500 and human user carry out mutual both software.According to some embodiments, such software includes but not limited to equipment device driver, operating system and user application such as SDK (Software Development Kit).Such computer-readable medium further comprises for implementing and comes the observability that precomputation looks from the district and the computer program of the embodiment of the method carrying out processing based on the content flow of increment PVS with Conservative Linear umbra event face.

Figure 56 is the block diagram of the processor 5600 of use in computing machine 5500.In one embodiment, processor 5600 comprises the CPU5680 that processes being stored in data among primary memory 5640 and/or the ROM5650 and instruction.CPU5680 also processes the information that is stored on disk 5610 or the CD-ROM5620.For example, CPU5680 is the X of IBM system from American I BM, and it adopts at least one from the Xenon processor of American I ntel or from the Opteron processor of U.S. AMD.Therefore, the instruction corresponding with the processing in the mobile device is stored among disk 5610, CD-ROM5620, primary memory 5640 or the ROM5650 any one.

In one embodiment, processor 5600 also comprise network interface 5675 as from the Intel Ethernet PRO network interface unit of the Intel Company of the U.S., display controller 5630 as from the NVIDIA GeForce GTX figure that is used for connecting display 5602 of U.S. NVIDIA company from adapter, display 5602 is such as Hewlett Packard HP L2445w LCD monitor.Processor 5600 also comprises for connecting keyboard 5695 and the positioning equipment 5685 I/O interface 5690 such as roller ball or mouse.According to some embodiments, Magnetic Disk Controller 5660 and disk 5610 interconnection, for example, hard disk drive or flash memory storage driver and CD-ROM5620 or the DVD driver with bus 5670, bus 5670 are ISA, EISA, VESA, PCI or the bus that similarly is used for all component of interlink server 5600.Because display 5602, keyboard 5695 and positioning equipment 5685 and show that general features and the function of control 5630, Magnetic Disk Controller 5660, network interface 5675 and I/O interface 5690 are known, so for for simplicity, the descriptions thereof are omitted.Certainly, its processing and hardware supplier and type are known in the art, for example, and from Freescale ColdFire, i.MX and the arm processor of U.S. Freescale company.

The example processor 5600 of Figure 56 is hardware platforms of computing equipment such as PC, and CPU580 is Intel Pentium processor, or the processor of other expectations as known in the art.Be stored among primary memory 5640, ROM5650, disk 5610 or the CD-ROM5620 computer-readable instruction on any one and be set to assembly or its combination of general purpose application program, backstage finger daemon or operating system, it is carried out in conjunction with CPU5680 with such as the operating systems such as Microsoft VISTA, UNIX, Solaris, LINUX, Apple MAC-OS and other system well known by persons skilled in the art.

Primary memory 5640 is random-access memory (ram), flash memories, eeprom memory etc., and ROM5650 is ROM (read-only memory) such as PROM etc.Because primary memory 5640 and ROM5650 are known, so for omitting for simplicity its other description.

Figure 57 shows the in one embodiment exemplary diagram of the relation between visibility events scrambler, visibility events server and the visibility events client.

In some embodiments, come the game database that comprises geometric figure, texture and other information or other modeling environments that are depicted as data 5710 are processed to produce the increment PVS data that are stored as visibility events data (5730) with Conservative Linear umbra event face.Described this processing of Figure 57 is undertaken by visibility events scrambler 5720.In one embodiment, off-line carries out this processing/coding to generate visibility events data 5730, and it is stored for subsequently.In some embodiments, visibility events scrambler 5720 comprises processor 5600 and carries out the processing shown in Fig. 1, Fig. 3, Fig. 4 A, Fig. 4 C, Fig. 5 A-Fig. 5 C, Fig. 6 A and Fig. 6 B.In other embodiment, the visibility events scrambler adopts the 3D trellis traversal of Figure 20 A and relevant figure to process to generate visibility events data 5730.

In some embodiments, transmit visibility events data 5730 in working time by the server unit that is labeled as " server ".In some embodiments, this server unit comprises the visibility events data 5730 of storing, and is before generated by the visibility events scrambler.Server unit can also be implemented visibility events demoder-server process 5740.In some embodiments, this visibility events server process can be implemented to be labeled as 5790 network interconnection and to pass to the client unit that is labeled as " client " so that the visibility events data communication device is crossed based on the server element of looking ahead of navigation.In some embodiments, the visibility events server can be implemented the packet control process based on consciousness discussed in conjunction with Figure 48 A, Figure 49, Figure 50 A, Figure 50 B and Figure 51.

In some embodiments, visibility events server 5740 is connected to game engine-server process 5750.Game engine-server process is used in the existing multi-player gaming usually, for example for the position that receives the player in the multi-player gaming with for these data are passed to client unit.By contrast, visibility events server 5740 transmits geometric figure, texture and other information progressively, and it comprises modeling environment, such as the visibility events data of looking ahead based on the user movement in the modeling environment in some embodiments.

Visibility events data 5730 are delivered to the client unit that is labeled as " client ", and it comprises visibility events demoder-client process 5780 in some embodiments.Visibility events client process 5780 receives visibility events data 5730.Process 5780 the visibility events data are processed into the PVS information that can play up.In some embodiments, bring in and carry out this and play up by being labeled as 5770 game engine client.

In some embodiments, demoder-client process 5780 receives the visibility events data of effectively being compressed by following methods: to having the profile contour that dynamically blocks efficiently and blocking that the border is identified and mark.In some embodiments, this effective compression is to process to carry out by contour identification and the mark described in conjunction with exemplary process diagram Figure 33 A, Figure 33 B, Figure 33 C and Figure 33 D.

In this embodiment, demoder-client process 5780 can be come according to the whole contour of identification in the single order silhouette edge of institute's mark of limited quantity (referring to, the exemplary process diagram of Figure 32 A and Figure 32 B) with the contour information that is included in the mark in the visibility events data of transmitting.The embodiment of demoder-client process can also working time from mark profile contour data generate whole block the border (referring to, Figure 34 A and Figure 34 B and Figure 35 A and Figure 35 B).

Use is according to these contour data of the side information generation of mark, in some embodiments, demoder-client process 5780 working time by from for example generating PVS(for the seed triangle of institute's mark of each coupling assembling of PVS or the increment PVS that generating to the traversal of contour, from the one or more sub-PVS of father PVS data) or increment PVS information (referring to the exemplary process diagram of Figure 36 and Figure 37).

In some embodiments, demoder-client process 5780 is connected with game engine client (5770).In some embodiments, the position of depending on the viewpoint position of user or other Agents, be delivered to demoder-client process or aforesaid demoder-client process go compress PVS that son generates in processing or increment PVS data and be committed to and play up.This plays up Microsoft DirectX or OpenGL-ES that the figure API that can adopt standard adopts such as the Playstation3 by Sony company.In some embodiments, these figures API is connected to graphic hardware by driver usually.

In some embodiments, demoder-client process is also obtained expression being used for or the information of the position of autonomous agents program in the modeling environment.In some embodiments, use two-way communication interconnection 5790 with this viewpoint position communication to demoder-server process.

Fig. 2 describes in detail: the structure of the identification of single order silhouette edge and single order Conservative Linear umbra event face

As previously mentioned, two kinds of visibility events faces that art methods is used to look from the district widely: for umbra volume integration method and the uncontinuity GRIDDING WITH WEIGHTED AVERAGE of area source.

In the umbra volume integration method, the visibility events face of constructing comprises umbra event face and the penumbra event face that intersects the border that forms corresponding shade volume.In simple situation, umbra event face is polygon (this paper is called umbra boundary polygon or UBP) and the border that forms polyhedral umbra volume.

It is the visibility events face of umbra and penumbra that the uncontinuity gridding method also adopts.In the uncontinuity gridding method, the visibility events face that is called as wedge does not intersect each other.Thereby the uncontinuity grid method does not for example produce clear and definite umbra volume.Alternatively, in the uncontinuity gridding method, wedge only intersects with the grid polygon.Then wedge-polygon intersects step, each wedge is used the 2D observability process to determine the visible part of crossing polygon segmentation.Polygonal these the visible segmentations of grid of intersecting form the uncontinuity border of this grid.This continuity border defines the same qualitative observability district (for example, umbra, non-penumbra etc.) on the polygonal mesh, and above-mentioned polygonal mesh can be determined after having constructed the continuity grid.

According to some embodiments, this method that the conservative observability of looking from the district is determined adopts the conservative linearization umbra visibility events face of constructing by the completely new approach of visibility events surface construction.

In a kind of embodiment of this method, these Conservative Linear umbra event faces intersect each other and intersect to form and UBP like the event Noodles that uses with the grid polygon in the shade volume method.

In the another kind of embodiment of this method, in fact these Conservative Linear umbra event faces intersect to form and wedge like the event Noodles that uses with the grid polygon in the uncontinuity gridding method.In the modification of the method, construct the observability figure (VM) that Conservative Linear is looked from the district according to these wedges.

The below is the general introduction of the first order modeling of observability propagation, and it is applied to two class single order visibility events faces: it can intersect wedge and UBP(to construct by wedge-wedge).

As it is apparent that from the analysis to prior art, the accurate visibility events face that limits the observability of looking from the district in the polyhedron environment often is quadric surface.These high terraces have proposed obvious calculating challenge, and it is so that the exploitation of the observability pre-computation methods of looking from district robust, actual is very difficult.

Embodiment comprises the method for the observability precomputation of looking from the district, the simplified model that it is propagated based on the observability of looking from the district in the polyhedron environment.This is called first order modeling.According to some embodiments, this model produced often be the plane, often be guard and continually for accurate visibility events face.Use test determines that a terrace is whether accurate and be used for measuring the maximum deviation of a terrace and precise results.Can come the single order method with higher-order method is that single order event face in coarse district carries out refinement.In some embodiments, higher-order method is the enforcement along reciprocal single order method: calculate the visible part of looking from the limit of looking the unit.

Different plane visibility events faces people 1992 such as () Heckbert with using in the uncontinuity gridding method guarantee that the visibility events face that conservative single order that this method adopts is looked from the district forms continuous umbra surface.These continuous umbra surfaces produce continuous uncontinuity net boundary, and it is divided into the uncontinuity grid in the district that looks visible district and look and be blocked from looking the unit from looking the unit.Thereby these districts form Conservative Linear umbra uncontinuity grid.Structure Conservative Linear umbra is disclosed from the method for the observability figure that looks the unit and look.Also pointed out to be used for the method for the conservative PVS that looks from the district that the observability figure that looks from the district according to correspondence derives.

According to some embodiments, suppose that polygonal mesh is that each limit accurately has two the polygonal envelope closed manifold of assembly triangular mesh (that is, by their common edge or one group of triangle of turning connection).In addition, suppose that vision area is the protruding unit of looking.In some embodiments, the single order observability determines that method does not need these hypothesis, but these hypothesis have strengthened the simplification of implementing.For example, polygonal mesh can be stream shape but not seal.In this case, each limit has one or two assembly triangle.

According to some embodiments, the observability model that single order is looked from the district is based on the conservative hypothesis of following simplification: if any element of polygonal mesh from vision area look (this paper be called look the unit) arbitrary portion as seen, then as seen it look from these all parts of looking the unit.This hypothesis causes the definition of the silhouette edge that single order looks from the district.

If the assembly polygon on limit of sharing polygonal mesh with respect to any summit of vision area just in the face of (as seen) and another assembly polygon with respect to all back sides, summit of this vision area to (invisible), then this limit is the silhouette edge that single order is looked from the district.The definition of single order silhouette edge also requires the assembly polygon not face with each other.

Compare the definition of the general silhouette edge of looking from the district (for example, employed by the people such as Dretakis and Nirenstein2005), this is a more limited definition.If for any summit of vision area, assembly polygon just in the face of and another assembly polygon back side pair, then the limit is the general silhouette edge of looking from the district.Differently state, if the limit is the silhouette edge of looking from point for any point in the vision area, then this limit is the general silhouette edge of looking from the district.

Following table is looked single order from the district silhouette edge and the general silhouette edge of looking from the district and the silhouette edge of looking from point compare.

Table III

The silhouette edge definition list

Single order is with the similarity of the definition of the silhouette edge of looking from point from the definition of the silhouette edge that the district looks: these two kinds of silhouette edges define the observability of looking from corresponding " district " that have for the viewpoint in the district that degenerates and the border between the complete invisibility.Significantly, if the assembly polygon with respect to protruding all back sides, summit of looking the unit pair, then it is looked the unit from this and looks invisible.As seen other assembly polygons that the definition of single order silhouette edge requires to share this limit look from this any point of looking on the unit.

Significantly, on any polygonal mesh, the silhouette edge that exists many silhouette edges of looking from the district than single order more generally to look from the district.Each single order silhouette edge be the general silhouette edge of looking from the district but be not each general silhouette edge of looking from the district be the single order silhouette edge.

According to determining the observability of looking from the district for the vision area of the stereoscopic unit of multiaspect in the present embodiment.

The visibility events face of looking from the district is incident on the silhouette edge of looking from the district.These visibility events faces of looking from the district can be penumbra or umbra.

According to some embodiments, as defined herein, the umbra visibility events face of looking from the district (also being called simply the umbra face) is to have the be not blocked directed face of side of side and look from the district that is blocked of looking from the district.Be arranged in looking from the district of umbra face be blocked on the side point from vision area look upper (or) any and institute be blocked a little.The point on the side that is not blocked that the district looks on the umbra face from vision area look upper () any point be not blocked (that is, visible).

The umbra visibility events face of looking from the district can be accurately or it can be to guard.

The umbra event face of looking from the district accurately in some embodiments, is by the quadric surface assembly and planar module forms and can be incident on any general silhouette edge of looking from the district.In order to determine which general silhouette edge of looking from the district supports accurate umbra event face, solve the exact solution of the visibility problem of looking from the district.As previously discussed, this is the computational problem of a difficulty, and it need to solve in high-dimensional space usually.

By contrast, the first order modeling that embodiment adopts observability to propagate, it has defined structure all is the plane and only is incident on the rotation of the conservative umbra event face on the single order silhouette edge and scans method.

In some embodiments, be positioned at point on the side that is blocked of conservative umbra event face and in fact look from vision area and be blocked, and in fact the point that does not block on the side that is positioned at conservative umbra event face may not be blocked or be blocked.Thereby, determine the observability of looking from the district with conservative umbra event face, for example, use conservatively from looking the observability mapping method of looking the unit, never underestimated from looking the unit geometric figure of looking visible, still may be over-evaluated.

Plane visibility events face (wedge) of adopting in the art methods of uncontinuity gridding is accurately, but they generally do not form continuous visibility events face.This is because accurate visibility events face generally is comprised of planar module and quadric surface assembly.Thereby the plane visibility events face of the art methods of uncontinuity gridding can not be used for determining this shadow zone.

By contrast, using the single order visibility events face of the method construct of present embodiment is the accurate or conservative inside that the continuous umbra event face that can use this method of the observability mapping of for example looking from the district is positioned at this shadow zone with definite what geometric figure that still is guaranteed to form.The penumbra event face of looking from the district is the directed visibility events face of incident on the general silhouette edge of looking from the district.On the side that is not blocked of penumbra event face, specific subarea or " aspect " of source vision area are visible.But on the side that is blocked of same penumbra event face, the identical subarea of vision area is blocked.The art methods of uncontinuity gridding is determined each assembly of the penumbra that throwed from area source by the polygonal mesh object with penumbra event face.

According to some embodiments, only adopt umbra event face to determine the observability of looking from the district.In one embodiment, these all umbra event faces are incident on the single order silhouette edge.In the embodiment that substitutes, can process adaptively refinement single order umbra event face with the umbra visibility events face of approximate exact more accurately by back projection.These refinements or " adjusting " visibility events face be such as single order umbra event face, plane; But they have reflected partial occlusion caused " high-order " the observability effect of looking from silhouette edge by vision area.Therefore these visibility events faces are called high-order visibility events face.In embodiment that this substitutes, these high-order visibility events faces (umbra) can " retract " non-single order, the general silhouette edge of looking from the district.

Table IV shows the type of the visibility events face that is incident on various types of silhouette edges and some feature of these visibility events faces.

Table IV

Be incident on the visibility events face on all types of silhouette edges

The basic sides of the first order modeling of propagating in observability shown in Fig. 2 A and Fig. 2 B.Description-embodiment at this instructions has partly provided follow-up details.

Fig. 2 A shows the figure that looks unit and two simple polygonal mesh A and B.

Fig. 2 A also shows the silhouette edge that two single orders are looked from looking the unit: it is subdivided into segmentation B10 and B1V limit A1 and limit B1().

The structure of the Conservative Linear umbra event face (CLUES) on these single order silhouette edges that is incident on is described now.It is similar that the umbra event face of constructing in the following discussion, and continuity grid wedge limit from the meaning of the observability of looking single the feature (be typically to support and look summit, unit or limit) of looking the unit from their.At the further part of this instructions, the observability figure (VM) that these wedges can be looked from looking the unit from its Conservative Linear of deriving PVS for structure is shown.

In some embodiments, limit from the whole single order umbra boundary polygon (UBP) of looking the observability of looking the unit and construct by intersecting clearly with corresponding single order wedge.Alternate embodiments in the subsequent section that structure and the purposes of UBP is depicted as this instructions.

Thereby, in Fig. 2 A and Fig. 2 B, illustrate with better simply wedge with single order UBP(at the single order wedge) structure in first step be identical.

Fig. 2 A shows and looks unit and mesh object A and mesh object B.In some embodiments, be the 3D cube with eight summits depending on the unit.For example, the unit of looking among Fig. 2 A is to have summit V ₁-V ₈Cube.In other embodiment, be the convex polyhedron of any expectation depending on the unit.The limit of grid A is marked as has summit A _{1_0}With summit A _{1_1}A1.The limit of grid B is marked as two segmentation: B1O and B1V.With respect to single order silhouette edge A1 because B1V be positioned at be formed on limit A1 with corresponding to looking summit, unit V ₈The support of the correspondence side of not blocking of looking event face " wedge 1 " between the SVV1 of summit, unit, look as seen so segmentation B1V looks summit, unit SVV1 from support.To this, because be incident on back side plane on the single order silhouette edge A1 along clockwise direction towards looking summit, unit V ₈Rotation is looked the summit, unit to determine corresponding support, so B1V is positioned at not blocking on the side of " wedge 1 ".Thereby, in some embodiments, be incident on back side plane on the single order silhouette edge is looked the event face that supports on the summit, unit towards the direction of looking summit, unit rotation the side of not blocking.The side of blocking of the event face that supports on the summit, unit is looked in the reverse direction indication of back side plane rotation.

With respect to single order silhouette edge A1, because B1O looks on the side that is blocked of the event face " wedge 1 " between the SVV1 of summit, unit with corresponding support being formed on limit A1, look and be blocked so segmentation B1O looks summit, unit SVV1 from support.

The single order visibility events face that is labeled as " wedge 1 " is at limit A1 and look in the supporting plane between the unit.Limit A1 and to look support polygon SP1 between the unit be summit A by limit A1 _{1_0}With summit A _{1_1}And the triangle (being labeled as SP1) that the summit, unit forms of looking that is labeled as SVV1.

According to some embodiments, the single order visibility events face " wedge 1 " on the A1 of limit of being incident on is by to being incident on the summit A of limit A1 _{1_0}With summit A _{1_1}On two limits of support polygon (SP1) of correspondence extend to form.This extends the summit A from A1 _{1_0}With summit A _{1_1}Place's beginning is along carrying out away from the direction semiinfinite ground of looking the unit.Article two, the ray of extension is connected to the summit A of limit A1 _{1_0}With summit A _{1_1}Be labeled as the semi-infinite umbra event face of " wedge 1 " with formation.Because " wedge 1 " in fact extends away from looking semiinfinite ground, unit, so only show the part of " wedge 1 " in Fig. 2 A.In some embodiments, the plane of event face is to be represented by 3D plane equation such as ax+by+cz.=0.

Therefore, in some embodiments, to be incident on the single order silhouette edge and to look (unit looks from looking) single order visibility events face on the summit, unit in order to form, at first structural configuration limit and look support polygon between the unit.This structure is similar with the rotary manipulation that silhouette edge is carried out, and above-mentioned rotary manipulation is along carrying out looking element characteristic (limit or summit) until run into support away from back side assembly polygon and towards the direction of looking the unit.In some embodiments, by forming wedge away from the non-silhouette edge of looking the unit and extend this support polygon.

As shown in fig. 2A, event face " wedge 1 " intersects with limit B1, B1 is divided into two segmentation B1V and B1O, B1V with respect to single order silhouette edge A1 from look element characteristic (depending on summit, unit SVV1) look into single order as seen, B1O looks from SVV1 with respect to single order silhouette edge A and is not that single order is as seen." wedge 1 " is comprised of segmentation B1O and B1V with single order silhouette edge B1() intersect at the some place that is labeled as CSV.

For purposes of illustration, suppose that now segmentation B1V is positioned at the side that is not blocked by all single order visibility events faces of Characteristics creation of the limit of grid A and " depending on the unit ".In this case, B1V is positioned at the outside (in the side that is not blocked) by the single order polyhedron polymerization umbra (PAU) that forms with single order wedge and grid polygon and with each other cross-shaped portion.Under these conditions, segmentation B1V from look the unit look into single order as seen.

If segmentation B1V from look the unit look into single order as seen, then under the conservative hypothesis of single order observability model, segmentation B1V is assumed to from looking any part of looking the unit as seen.Thereby, look the rotary manipulation that generates support polygon (SP2) between the summit, unit by aforesaid at segmentation B1V and the support that is labeled as SVV2, construct the single order visibility events face that is incident on the segmentation B1V.As shown in fig. 2A, support polygon SP2 is by the summit of segmentation B1V and looks summit, unit V ₃(SVV2) limit.Adopt again the method for aforesaid extension support polygon.Result's the single order visibility events face that is incident on the BV1 is marked as " wedge 2 ".

Because corresponding support is looked summit, unit SVV1 and is in fact looked as seen from the single order silhouette edge A1 that supports in this case, so " wedge 1 " is the accurate visibility events face that is incident on the A1 of limit.

" wedge 2 " is not the accurate observability plane of passing B1V, this is because the conservative hypothesis of single order observability model is violated in very special mode: it is sightless that corresponding support looks in fact summit, unit SVV2 look from the single order silhouette edge B1V that is supported, when observing from this limit, SVV2 is blocked.

Use is looked the degree of accuracy that 2D observability test that the observability of looking from silhouette edge on summit, unit tests can determine to be incident on any single order visibility events face (for example, wedge) on this silhouette edge to support.In some embodiments, to look element characteristic be the summit if support, and then this is an observability test of looking from point, its with the support from corresponding of single order silhouette edge is looked the test that the observability of looking on summit, unit (SVV) carries out and is equal to.In some embodiments, accurate observability sexual behavior part face is supported in the SVV from correspondence of the single order silhouette edge segmentation of looking visible, and out of true/conservative visibility events face is supported in the SVV from correspondence of the single order silhouette edge segmentation that is blocked of looking.

Look element sides in particular cases with supporting silhouette edge is parallel, to need the special observability of looking from limit test.In the subsequent section of this instructions, this is elaborated.

Embodiment also comprises the method for the precision that increases out of true visibility events face.In the method, for each segmentation of the support out of true wedge of single order silhouette edge, be identified as for the visible support of this segmentation and look summit, unit (VSVV) looking point on the face of unit.In fact VSVV looks visible and forms support polygon with this segmentation from the silhouette edge of correspondence.

According to some embodiments, by back projection definite VSVV: look from the district visibility events face of (looking from the limit in this case) of the single order that uses silhouette edge oppositely to look unit projection by the polygonal mesh object from line source as line source and structure.These single order wedges and grid polygon and with the cross-shaped portion of looking the unit comprise from silhouette edge look, about looking the observability figure of unit.This observability figure comprise look the unit from the silhouette edge assembly of looking visible.VSVV is the support summit of these visible assemblies.

By coarse silhouette edge section rotary is gone to wedge that corresponding SVV constructs be regulate or " high-order " visibility events face.These high-order visibility events faces have reflected the partial occlusion effect of looking from silhouette edge of looking unit (source), a kind of effect that can not explain with the simple conservative first order modeling that observability is propagated.

Fig. 2 B shows the result who B1V is considered as the back projection processing of line source.Construct the summit A that is incident on limit A1 according to the segmentation B1V that is regarded as line source _{1_0}With summit A _{1_1}On be labeled as the wedge of " wedge _ back of the body ".Note, limit A1 is the single order silhouette edge with respect to source region B1V.In this example, the district below " wedge _ back of the body " is the side that is not blocked of " wedge _ back of the body ", its expression look the unit from the B1V part of looking visible.

Support polygon between B1V and the A1 is the triangle with summit VB of limit A1 and limit B1V.Corresponding wedge " wedge _ back of the body " has created the new visible contour that comprises summit VSVV of looking the unit with to look the unit crossing.

In some embodiments, structure back projection wedge has used the identification of single order silhouette edge and rotation and has scanned the method that wedge is constructed such as the processing of " wedge _ back of the body ", as described in some embodiments in this manual.When looking the unit and use these methods as vision area, result's single order wedge is away from looking the unit extends and intersect with polygonal mesh, and polygonal mesh is divided into from looking the unit single order part that part and single order be blocked of looking visible.By contrast, if these methods are applied to back projection, then corresponding wedge extend away from the single order silhouette edge (such as B1V) as the straight line vision area and with look the unit and intersect, become from the look visible part of single order and from the silhouette edge part that single order is blocked of looking of silhouette edge looking dividing elements.Define new elementary boundary or the contour (be also referred to as visible support and look the unit silhouette edge) as seen looked depending on this division of unit, as seen it look conservatively from the single order silhouette edge as the straight line vision area.Then the summit of this contour is tested to determine on which summit to be for will be on the support summit of the high-order wedge of constructing as the single order silhouette edge of straight line vision area (the visible support looked the summit, unit).

Be generated as limit B1V and the visible support polygon " SP_ is high " of looking the support summit between the unit contour is constructed " adjusting " or high-order wedge by the rotation from BV1 to VSVV.

The non-silhouette edge of high-order support polygon " SP_ is high " passes the summit of B1V and extends, and as previously mentioned, forms high-order wedge " wedge _ height ".

Therefore, in order to construct from looking the high-order visibility events face on the single order silhouette edge of looking the unit, along use in the other direction single order method that observability propagates with determine to look the unit from the silhouette edge part of looking visible.

Shown such as the subsequent section at this instructions, can estimate adaptively based on error the single order silhouette edge that supports coarse visibility events face is segmented.Can be at the segmentation structure high-order wedge of estimating to instruct segmentation by these errors so that the result to be the sectional plan of corresponding accurate secondary event face approximate.In addition, this method of applied single order observability has realized the quadric new method of structure, and it is guaranteed that the face of constructing is conservatively underestimated and blocks, even polymerization is in accurate result.

For purposes of illustration, the visibility events face of single type has been adopted in the aforementioned theoretical property introduction of single order observability.Between (depending on the unit) summit, source and silhouette edge, form the visibility events face of the type.The visibility events face of the type is called the SV-ME wedge.The visibility events face of another kind of type is used for constructing the continuous visibility events face of looking from the district that is incident on the non-convex polygon grid.The visibility events face of the type is by looking that (source) limit, unit and grid vertex form and be called the SE-MV wedge, in other parts of this instructions it is discussed in detail.

In a word, method is constructed the single order wedge to use simple single order according to the single order silhouette edge " to rotate-look the unit ".The single order wedge can with grid polygon and other wedges intersect form continuous in looking the observability figure that looks the unit or continuous single order PAU.The two data structure is conservatively underestimated from looking blocking that the unit looks.Embodiment comprise derive from single order observability figure or single order PAU conservative from looking the enforcement of the PVS that looks the unit.

Can determine to look by using single order observability method the back projection of the visible part of silhouette edge of unit and process to construct high-order visibility events face.

Above detailed description introduced the single order observability model that observability propagates and the overview that is used for the certain methods of structure single order visibility events face.Also will provide in further detail in this manual the details about the identification of single order silhouette edge and single order wedge structure.

The description that Fig. 1 to Fig. 6 is detailed: the first embodiment that adopts the conservative linearization umbra event face of constructing with rotation and the method for scanning.

A kind of embodiment comprises the method for constructing conservative linearization observability figure based on the first order modeling of the simplification of the propagation of the observability in the polyhedron environment.Described in the embodiment, single order observability model is based on following conservative hypothesis such as the front: if the silhouette edge of polygonal mesh from the arbitrary portion looking the unit and look as seen, then as seen this silhouette edge from looking all parts of looking the unit.According to the embodiment of this model, silhouette edge (being called the single order silhouette edge) is limited to and has an assembly polygon of just facing with respect to all summits of looking the unit and have polygonal those triangular mesh limits of another assembly right with respect at least one back side, summit of looking the unit.In addition, become the single order silhouette edge, the assembly polygon is the back side pair relative to each other also.

This model also causes following methods: look the unit silhouette edge and scan (deriving the SE-MV wedge from the support triangle shape of scanning) and form the linearization umbra event face (be called as CLUES, or be called the single order wedge or be called simply wedge) that single order is guarded by the rotation from (single order) silhouette edge to the summit of looking the unit (deriving the SV-ME wedge from the support polygon of rotation) or by spreading all over from (single order) inside corner profile summit.The method is also used in the silhouette edge that supports and support and is looked the parallel SE-ME wedge that in particular cases generates of element sides.This single order embodiment always produces conservative umbra border, and in some cases, it is accurate umbra border.

The high-order model that other embodiments are propagated based on observability in the polyhedron environment.If this model does not have the hypothesis silhouette edge from any part of looking the unit and looking visible then as seen it from looking all parts of looking the unit.More properly, this model explanation look the part of looking and being blocked from silhouette edge of unit.High-order model forms the basis of alternative embodiment, and above-mentioned alternative embodiment can produce accurate umbra border more approximate in the coarse situation of single order method.

The single order embodiment of this method is at first described.

Fig. 1: the structure of single order wedge

Fig. 1 show disclose use rotation and the method for scanning by the process flow diagram of the overall tissue of polygonal mesh object and the protruding structure of looking the single order wedge that the unit forms.According to some embodiments, treatment scheme wherein checks the polygon of polygonal mesh object individually for the single order silhouette edge from step 110.The method of identification single order silhouette edge is disclosed in Fig. 3 in detail.Figure 20 illustrates and disclose the embodiment that checks polygonal order, it shows the algorithm of implementing strict vertical order.

Treatment scheme marches to step 112, and to look element sides parallel with support with the single order silhouette edge determining to run in step 110.

To look the summit, unit not parallel with support if determine the single order silhouette edge in decision steps 112, and then treatment scheme marches to step 114 with the structural configuration limit and looks support polygon between the unit.Fig. 4 A shows the details of this structure.

Treatment scheme marches to step 116 with by the polygonal certain edges thereof of the rotary support of the correspondence on the summit that is incident on the single order silhouette edge being extended to construct the SV-ME wedge that is incident on the single order silhouette edge.The additional detail of step 120 is disclosed in Fig. 6 A.

If the single order silhouette edge is parallel with support profiles, then treatment scheme marches to step 118 from step 112.

In step 118, as previously mentioned, by constructing the support quadrilateral that is called as SE-ME quadrilateral (source limit-Grid Edge) to the rotation of looking the unit from looking element sides.

Treatment scheme marches to step 120 with by extending to construct with SE-ME and support SE-ME wedge corresponding to quadrilateral being supported line segment that tetragonal diagonal line summit forms by SE-ME.The limit of SE-ME wedge is by supported silhouette edge with by passing the silhouette edge summit and extend two lines that the diagonal line line segment forms and form away from looking the unit.

Treatment scheme marches to decision steps 125 to determine whether adjacent silhouette edge forms the external corners of single order profile contour from step 116 or step 120.In some embodiments, use and make this for the simple test of the relative direction of adjacent silhouette edge and determine.Borderline each limit that is positioned at polygonal mesh has following natural direction: a normal on this limit is inner towards polyhedron towards the polyhedron outside (outside normal) and relative normal.If two outward normals for adjacent silhouette edge are faced away from each other, the profile summit of then sharing is the inside corner of this profile contour.Otherwise the profile summit of sharing forms external corners.

If determine that in step 125 adjacent silhouette edge has formed the external corners of profile contour, then treatment scheme marches to step 140 and intersects each other with the wedge that will be incident on the adjacent silhouette edge.In some embodiments, if adjacent SV-ME wedge is to look summit, unit (SVV) by rotation to identical support to generate, then they accurately intersect at the common edge place.Otherwise adjacent SV-ME wedge intersects each other in their polygon inside and makes clear and definite polygon-polygon cross-shaped portion and determine.In any situation, the SV-ME wedge that intersects produces the continuous umbra event face of the part of crossing over the single order profile contour that is formed by two supported silhouette edges.In some embodiments, adjacent SV-ME wedge does not intersect.In these embodiments, step 140 is optional.The SV-ME wedge that does not intersect with adjacent SV-ME wedge still can intersect with the grid polygon, and result's wedge-grid polygon cross-shaped portion is tested to determine that it is whether from looking the border of blocking that the unit looks.Provided the other discussion of the adjacent S V-ME wedge that intersects in conjunction with Fig. 7 D4 and Fig. 7 D5.

If determine that in decision steps 125 adjacent silhouette edge does not form the external corners of profile contour, then treatment scheme marches to step 130 from step 125.This situation is corresponding with the inside corner of single order profile contour.

In some embodiments, such inside corner by forming by two continuous silhouette edges of summit is simple profile summit.The first order modeling that uses observability to propagate, inside corner can also be formed on the assembly silhouette edge and not share on the compound-contoured contour on the summit in the primary flow shape grid.These summits are called as compound-contoured summit (CSV), and they are corresponding to the t summit of looking from the district of stream shape, and at the subsequent section of this instructions it are discussed in detail.

In step 130, on inside corner grid profile summit with look forming one or more supports with respect to this inside corner grid profile summit between for some limit of the silhouette edge of looking from point and scanning triangle of unit.The additional detail of this processing is disclosed in Fig. 5 A and Fig. 5 B.

Treatment scheme marches to step 135, scans the next SE-MV wedge according to scanning triangle generation correspondence in leg-of-mutton limit by passing the extension of inside corner grid profile summit.The detail of this processing is disclosed in Fig. 6 B.

The alternate embodiments of constructing the set of single order wedge with diverse ways can be arranged.For example, in a kind of alternate embodiments, can construct the whole conservative support package of looking between unit and the polygonal mesh object, and elect the single order wedge as the conservative polygonal subset of support package.

Fig. 3 shows the details of the step 110 among Fig. 1, the identification of single order silhouette edge.According to some embodiments, enter processing shown in Figure 3 at step 110 place of Fig. 1.In some embodiments, the processing of identification single order silhouette edge starts from identifying the polygonal step 310 of assembly on this limit.In some embodiments, by polygonal mesh being stored as linked data structure such as wing limit data structure makes things convenient for this processing, in above-mentioned data structure, store for the polygonal reference of the assembly on each limit.In some embodiments, the data structure with any expectation represents polygonal mesh.In a kind of enforcement, polygonal mesh is closed stream shape, and wherein each limit is accurately shared by two assembly polygons.

Treatment scheme marches to an assembly polygon that 315 pairs of decision steps are called as polygon B or PB and tests to determine that this assembly is with respect to the back side pair whether, all summits of looking the unit.In this case, all summits depending on the unit can be positioned on the rear side that comprises this polygonal plane of assembly.

Look the not back side pair, summit, unit if in decision steps 315, determine PB with respect to all, then process from step 315 and march to step 320 so that another assembly polygon that is called as PA is tested, described in step 315.

If determine PA with respect to all back sides, summit of looking the unit pair in decision steps 320, then treatment scheme marches to step 325 to determine that assembly triangle PB looks the summit, unit with respect at least one and whether just faces.

If determine that in decision steps 325 PB looks the summit, unit with respect at least one and just faces, then process marching to 330 couples of PA of decision steps and PB and test to determine their relative to each other back side pair whether.

If determine the relative to each other back side pair of PA and PB in decision steps 330, then treatment scheme marches to step 335, and the limit of wherein testing is the single order silhouette edge.

If determine the relative to each other back side pair not of PA and PB in decision steps 330, then treatment scheme marches to step 355, and it returns the limit of testing is not the result of single order silhouette edge.

If determine PB with respect to all back sides, summit of looking the unit pair in step 315, then treatment scheme marches to step 340 to determine that PA looks the summit, unit with respect at least one and whether just faces.If PA looks the summit, unit with respect at least one and just faces, then treatment scheme marches to step 345 determining the relative to each other back side pair whether of PA and PB, as described on the function in step 330.

If PA and PB be the back side pair relative to each other, then treatment scheme marches to step 350, and it returns the result that the limit of testing is the single order silhouette edge.If PA and PB be the back side pair not relative to each other, then treatment scheme marches to step 355 from step 345.If PA looks the summit, unit with respect at least one and just do not face, then the treatment scheme row marches to 355 from step 340.If any test in step 320,325,330,340 or 345 is failure all, then this Grid Edge is not the single order silhouette edge, as indicated in step 355.

The building method of Fig. 4 A:SV-ME and SE-ME support polygon

Fig. 4 A shows the process flow diagram that structure is incident on the method for the SV-ME support polygon on (single order) grid silhouette edge.Fig. 4 A has provided the additional detail in the processing shown in the step 116.According to some embodiments, from the step 116 of Fig. 1, enter the processing shown in Fig. 4 A.

In some embodiments, when running into the single order limit of polygonal mesh, the processing of structure support polygon is from step 410.Although may use potentially the high-order silhouette edge at other embodiments, this is a single order silhouette edge in the present embodiment.

Treatment scheme marches to step 415 with the single order silhouette edge and look " support _ angle " between the unit and be set to " maximum " value (for example, 180 degree).According to some embodiments, support angle is defined as: when beginning to rotate through the plane of single order silhouette edge from the polygonal plane of back side assembly and towards looking finite element rotation (along the general direction of the polygonal normal of back side assembly) until formed angle when running into the first summit of looking the unit or limit.Plane of rotation is silhouette edge and the plane of looking the support polygon between the unit with looking the position that the unit contacts.During this rotation the angle of process be called as support angle or the anglec of rotation, and between the polygonal plane of assembly, the back side of supporting plane and silhouette edge, this angle is measured.Look summit, unit or limit (if support polygon is the SE-ME type) of causing the minimum anglec of rotation are to support summit or support edge.

The remainder of Fig. 4 A shows identification and supports the processing of looking unit and structure support polygon.Treatment scheme marches to step 420 " summit " is set to look the first summit of unit.In embodiment, " summit " is to determine that whether this candidate vertices is as supporting the summit for the candidate vertices of testing.Treatment scheme marches to step 425 with the triangle between the structure grid silhouette edge " limit " and " summit ".Treatment scheme marches to step 430 with being used for measuring angle between the viewable side on the polygonal plane of assembly, the back side of the plane of " triangle " and this silhouette edge in the standard method of the angle between the intersection place on the plane measurement plane.Treatment scheme marches to step 435 and compares with the currency that will be somebody's turn to do " angle " and " support _ angle ".If " angle " less than the currency of " support _ angle ", then treatment scheme marches to step 440 so that " support _ angle " is set to " angle ".Treatment scheme marches to step 445 so that " support _ summit " is set to current " summit ".

Treatment scheme marches to step 450, and wherein support polygon is set to by silhouette edge and supports the triangle that the summit forms.

Treatment scheme marches to step 455 has the untreated summit, unit of looking to determine whether residue.If determine not remain the untreated summit, unit of looking in decision steps 455, then treatment scheme marches to step 460, wherein support polygon is exported.

Look the summit, unit if determine in decision steps 455 that residue is untreated, then treatment scheme marches to step 475, wherein select for the treatment of the next one look the summit, unit.

If determine that in decision steps 435 measured " angle " (anglec of rotation) is not less than current " support _ angle ", then treatment scheme marches to step 465 to determine whether the anglec of rotation (" angle ") equals the currency of " support _ angle ".If this condition is set up, two summits then looking the unit with form identical angle corresponding to silhouette edge, corresponding and SE-ME support polygon, and treatment scheme marches to step 470 and looks the summit, unit and look quadrilateral (SE-ME support polygon) between the element sides to arrange two.

The in particular cases structure quadrilateral support polygon in step 470 just that only equates with two support angles of looking between the summit, unit at silhouette edge.For the protruding unit of looking of supposition in the present embodiment, this only appears at when two supports are looked the summit, unit and are positioned on the limit parallel with the grid silhouette edge of looking the unit.In this case, not to be determined by common observability triangle of looking from point but alternatively determined by the observability quadrilateral of looking from segmentation from the observability of looking " leap " silhouette edge of looking the unit.

Other embodiments of differently processing these special circumstances can be arranged, for example, be incident on parallel support by structure and look two support triangle shapes on the element sides and scan triangle.Use the method, the adjacent UBP of result's correspondence can not intersect at the place, limit, but alternatively, their can be overlapping on their plane, thereby causes the regional degeneration of the polyhedron umbra volume that surrounds.This method of identification quadrilateral support polygon has been avoided this degeneration in the later step.

No matter candidate's support polygon is triangle or quadrilateral, treatment scheme all marches to step 455 to determine whether to remain any untreated summit, as mentioned above from step 470.If residue is looked the summit, unit, then the treatment scheme row is back to step 475, wherein selects the next one to look the summit, unit.Then, abovementioned steps is followed in processing.

At final step 460 place, process the output support polygon, it is by the grid silhouette edge and looks the triangle that the summit of unit forms, or is formed on the grid silhouette edge and looks quadrilateral between the element sides.

The alternate embodiments that the method for structure SV-ME support polygon can be arranged.In a kind of alternative embodiment, " support _ summit " corresponding with single order silhouette edge is limited to directly those that link to each other with " support _ summit " for adjacent single order silhouette edge and looks the summit, unit, wherein the external corners (protruding feature) of adjacent limit formation grid.The method of using in the art methods of classics of the method and the method for dividing and ruling of structure convex closure in 3D is similar.In this application, be that the acceleration that very simple polyhedron and the method are afforded is very limited depending on the unit.

Fig. 4 B shows mesh object M1 and looks the unit.In the same object shown in Fig. 7 A and Fig. 7 B1 depending on unit and polygonal mesh M1.In Fig. 4 B, viewpoint is between the viewpoint of Fig. 7 A and Fig. 7 B1.The single order silhouette edge that is labeled as B also appears among this three width of cloth figure.The direction of observation of Fig. 4 B is in close proximity to parallel edges B.Therefore, the limit is almost regarded limit B a little as on the most front ground (edge-on).Summit with polygonal mesh M1 in Fig. 4 B and Fig. 7 B1 is considered as summit V3.

Two candidate's support polygons are shown " candidate SP1 " and " candidate SP2 ".Come for single order silhouette edge B identification candidate support polygon by limit B with the triangle that the summit of looking the unit forms by structure.The angle that the polygonal plane of the back side assembly of the plane of this support polygon and shared limit B forms is measured.This angle corresponding in the step 425 of Fig. 4 A, determine and step 435 in same figure and step 465 in the variable " support _ angle " that uses.In the example shown in Fig. 4 B, the back side assembly polygon of single order silhouette edge B is the triangle that is formed by limit B and summit V3.

In this example, the angle that is formed by " candidate SP1 " (corresponding to summit V4) represents with the broken arcs that is labeled as " angle-1 ".

In this example, the angle that is formed by " candidate SP2 " (corresponding to summit V8) represents with the broken arcs that is labeled as " angle-2 ".

According to these two arcs, significantly, " angle-1 " is less than " angle-2 ".According to the exemplary process diagram of Fig. 4 A, candidate SP1 can be remained the candidate of the support polygon of the reality on the single order profile.If the processing shown in the exemplary process diagram in Fig. 4 A is tested all summits of " depending on the unit ", can find that then summit V4 produces the support polygon (" candidate SP1 ") that provides the minimum support angle.In Fig. 7 C1, " candidate SP1 " is shown the support polygon SPB of reality.

Can adopt the measurement of angle of standard to determine to comprise the angle of the cross product between the normal line vector on plane of backfacing polygon and candidate's support polygon.

Fig. 4 C shows be used to determining to be formed on the single order silhouette edge and looking the polygon process flow diagram of the test of support polygon whether between the summit, unit.

Following alternate embodiments can be arranged: identify the SV-ME support polygon by " side (sidedness) orientation " (with respect to the inside of polygonal mesh) of consideration candidate support polygon and the orientation with respect to looking the summit, unit of candidate's support polygon.

In one embodiment, the grid polygon all is presumed to be " outside " polygon with following normal line vector: normal line vector is oriented to " inside " away from the included district of polygonal mesh partly.In such embodiment, all grid polygons of polygonal mesh as one man have this identical " side " orientation.

Polygon is a kind of planar structure, and it can have two sides corresponding to the both sides that comprise this polygonal plane.Illustrative embodiments is included as stream shape or closed polygonal mesh.The spatial volume that stream shape grid will be embedded with it is divided into inside and outside.In calculating graphics, it is very useful using with dirty shape grid: the polygonal normal line vector of each in this grid is oriented partly towards the inside away from this closed volume.This can be called as polygonal " outside " side.Opposition side can be called as polygonal " outside " side.If all polygons all have this consistent side sexual orientation in grid, then should can exist never polygonal private side to look as seen from the outside.

In the exemplary embodiment, can namely share the polygonal vertex order on limit and establish the polygon of grid and have identical side sexual orientation by checking adjacent polygon.(referring to Schneider (2003) Schneider, Philip J., Eberely, David H., " Geometric Tools for Computer Graphics " Morgan Kaufmann2003pp.342-345, its full content merges among the application by reference).If F ₀And F ₁To share by two summit V ₁And V ₃Two adjacent polygons on the limit that forms.For polygon F ₀If, summit V ₁And V ₃With V ₃Follow V ₁Ordering occur, then they are at polygon F ₁In must be with V ₁Follow V ₃Ordering occur.The limit of sharing have this ordering adjacent polygon can be called and be had consistent summit ordering.Polygon with consistent summit ordering has identical side sexual orientation.The summit ordering has reflected the order for each triangle storage summit.Having defined its cross product for triangle with the summit of this identical order access is coefficient A, the B of plane equation, vector (triangle edges) or this leg-of-mutton normal line vector of C.In some embodiments, all mesh triangles shapes have consistent summit ordering and all mesh triangles shapes can have sensing away from the normal line vector of the direction of the inside of grid, that is, they all are the triangles towards the outside.Sort (referring to MakeConsistent procedure of Schneider (2003), pp345) in the inconsistent summit that embodiment can adopt known algorithm to identify and repair in the polygonal mesh before processing.

Fig. 4 D1 shows two adjacent polygon F that polygon has consistent summit ordering ₀And F ₁Exemplary diagram.Note, for polygon F ₀, be with V ₁To V ₃Order visit shared limit, and for adjacent polygon F1, with V ₃To V ₁Order visit identical shared limit, therefore satisfy the definition of consistent ordering.Employing right hand rule tradition, two polygonal normals point to the outside of the plane of delineation.

Fig. 4 D2 shows two adjacent polygon F that polygon does not have the ordering of consistent summit ₀And F ₁Exemplary diagram.

In one embodiment, be formed on for the candidate SV-ME support polygon of single order silhouette edge and look between summit, unit and the single order silhouette edge.Provided the candidate's support polygon with side sexual orientation identical with the polygonal side sexual orientation of the back side grid of shared single order silhouette edge.(use this consistent side sexual orientation, for example, the people who passes through the single order silhouette edge walking on back side grid polygonal " outside " face can run into candidate's support polygon " outside " face) then check with respect to looking the orientation of summit, unit to the plane of each candidate's support polygon.If the plane of candidate's support polygon is looked the summit, unit with respect to each and just do not faced, the summit, unit of looking that then forms candidate's support polygon is to support to look the summit, unit, and candidate's support polygon is support polygon.

According to some embodiments, that adopts looks summit, unit (that is, admit to support look the summit, unit be non-front) with respect to what look the definition of just facing on the summit, unit got rid of the plane that is arranged in candidate's support polygon.The embodiment that substitutes can adopt the modification of the definition in the back side and front to look the summit, unit with respect to each and just do not face to determine candidate's support polygon.In at least a illustrative embodiments, this test comprises confirms that candidate's support polygon looks the back side, summit, unit pair with respect to each, wherein the definition on the plane right with respect to the back side, summit comprise the summit that is arranged in this plane (that is, admit to support look the summit, unit with respect to the support polygon back side to).

According to some embodiments, from step 480, enter the processing shown in Fig. 4 C.In step 408, at the single order silhouette edge and look between the summit, unit (V) and to form candidate's support polygon.

Treatment scheme marches to step 485 and is arranged to identical with the polygonal side sexual orientation of the back side assembly of sharing the single order silhouette edge with the side sexual orientation of candidate's support polygon that will form in step 480.

Treatment scheme marches to step 487 to determine whether candidate's support polygon is just faced with respect to each summit of looking in the summit, unit.Look the summit, unit with respect to each and just do not face if determine candidate's support polygon in decision steps 487, then treatment scheme marches to step 481 and is identified as and looks summit, unit (V) and support polygon is identified as candidate's support polygon support is looked the summit, unit.

Just face with respect to any summit, unit of looking if determine candidate's support polygon in decision steps 487, then treatment scheme marches to step 489 and is identified as non-support and looks the summit, unit and this candidate's support polygon is identified as non-support polygon will look summit, unit (V).

The test that is illustrated by the exemplary flow of Fig. 4 C can also be used for identification SE-ME type support polygon.

Fig. 5 A and Fig. 5 B: show the process flow diagram that structure SE-MV supports (scanning) leg-of-mutton method

Fig. 5 A and Fig. 5 B comprise that showing structure is incident on SE-MV on the inside corner grid profile summit and supports and to scan the process flow diagram of leg-of-mutton method.This is the additional detail of the step 130 of Fig. 1.According to some embodiments, enter the processing shown in Fig. 5 A and Fig. 5 B in the step 130 from Fig. 1.

In some embodiments, when the inside corner of the single order profile contour that runs into polygonal mesh, structure SE-MV supports and scans leg-of-mutton processing and begin from step 510.This inside corner can be formed by the simple single order profile contour on the shared summit of two single order silhouette edges.If the normal of the silhouette edge of this cross-shaped portion of this formation (with being assumed to towards the normal direction away from the direction of the polygonal inside of assembly of silhouette edge) faces with each other, then this cross-shaped portion is the inside corner summit.

Alternately, the inside corner summit can be the summit of the compound-contoured contour that forms of the cross-shaped portion by wedge and single order silhouette edge.Under latter event, inside corner profile grid profile summit is called compound-contoured summit (CSV).

Treatment scheme marches to step 515 and looks summit, unit (SVV) for example to identify for the support of one of silhouette edge that forms the summit with the disclosed processing of Fig. 4 A.The sign on this summit is stored as variable " SVV_ begins ".Treatment scheme marches to step 520, is variable " SVV_ end " for the processing of another limit repeating step 515 of inside corner and with result store wherein.

If arbitrary support polygon of inside corner is quadrilateral (generating in the step 470 of Fig. 4 A), then support polygon has two SVV.In particular cases this, must in step 515 and 520, carefully select initial in this chain to look summit, unit (" SVV_ begins ") or stop looking summit, unit (SVV_ ends), this is to be that the other end from this chain removes farthest because of this summit.

Treatment scheme marches to step 525, and wherein variable " current _ polygon " is set to identification and looks corresponding support polygon between the limit of supporting of summit, unit " SVV_ begins " and polygonal mesh.

Treatment scheme marches to step 530, wherein will be configured to look summit, unit " SVV_ begins " and be stored as variable CVV for finally occurring in the starting point that unit profile contour scans of looking of looking between summit, unit " SVV_ begins " and " the SVV_ end ", it have preserved the current summit of scanning.

Treatment scheme marches to 535 couples of CVV of decision steps and compares to determine whether stop scanning with " SVV_ end ".

If definite processing identical with last summit (" SVV_ end ") in scanning when summit, forward sight unit (CVV) in decision steps 535, then treatment scheme marches to step 540 and stops.If two limits of inside corner have the identical point of looking on the unit, then corresponding SV-ME wedge intersects and does not exist the scan triangle corresponding with the inside corner summit along common edge.Can this situation of identification in the original execution of step 535 and can stop scanning and do not produce and scan triangle.

If determine that in decision steps 535 CVV is not " SVV_ end ", then treatment scheme marches to step 545 variable " current _ angle " is arranged to maximal value.

Treatment scheme marches to step 550, wherein select to share to look first of summit, unit CVV and look element sides and come mark with variable " limit ".

Treatment scheme marches to decision steps 555 to determine that limit " limit " is whether with respect to (looking from point) silhouette edge of inside corner grid profile summit MV.

If determine that in decision steps 555 " limit " is the silhouette edge of looking from MV, then treatment scheme marches to step 560 to form triangle between a MV and limit " limit ".This triangle is that MV and the candidate that looks between the unit scan triangle, but must with its with share identical look element sides other scan the triangle candidate and compare.

Treatment scheme marches to 565, wherein begins comparison these other and scans the triangle candidate.To this, the current angle of scanning between triangle candidate " triangle " and " current _ polygon " (support polygon) that is incident on the MV is measured.This value is stored in the variable " angle ".Because " triangle " and " current _ polygon " shares common edge, so can adopt this angle to come to measure this angle at this place, limit for the convention of each polygonal angle between the side that is blocked.The side that is blocked of support polygon is a side that is connected to the polygonal inside of grid at the silhouette edge place.It is a side that is connected to the polygonal inside of grid at summit MV place that the candidate scans the leg-of-mutton side that is blocked.This angle is stored in the variable " angle ".

Following alternate embodiments can be arranged: the orientation with respect to adjacent wedge of scanning triangle and the SE-MV wedge of correspondence is checked.All wedges all are the directed face with " as seen " side and " invisible " side.For the SE-MV wedge, viewable side is the side (do not block owing to not being exceeded the grid polygon of corresponding single order silhouette edge, so be visible in this side) that is not blocked.For the SV-ME wedge, viewable side is " involved " side (owing to looking in the unit when passing and exceeding to be comprised in when corresponding inside corner single order profile summit is observed, so be visible).

In one embodiment, look element sides according to MV and construct " scanning _ triangle ", MV look element sides produce have with adjacent SV-ME wedge block direction consistent and with adjacent S V-ME wedge comprise direction consistent " scanning _ triangle " of comprising direction.The SV-ME wedge that does not have this consistent direction is not facilitated continuous, conservative linearization umbra event face.

The opposite orientation of the orientation of SV-ME wedge and corresponding SV-ME support polygon.Because the limit of SV-ME support polygon is passed inside corner single order profile summit and effectively " is projected " and has formed corresponding SV-ME wedge, so this reversing has occured.(for example, have look the unit can produce with this SE-MV support polygon that comprises axle between the inside corner single order profile summit that " is lower than " specific SE-MV support polygon along negative Y-direction have it along the corresponding SE-MV wedge of positive Y-direction " involved " or viewable side).

Treatment scheme marches to decision steps 570 to determine that this angle (" angle ") is whether less than the currency of " current _ angle ".

If determine the currency of " angle " less than the value of " current _ angle " in decision steps 570, then " triangle " is that the candidate scans triangle and treatment scheme marches to processing 5-1, and its step 580 from Fig. 5 B begins.

In step 580, variable " current _ angle " is set to the value of " angle ".

Treatment scheme marches to step 585 and is arranged to refer to limit " limit " so that variable " is scanned _ limit ".

Treatment scheme marches to step 590 and is arranged to refer to triangle " triangle " so that variable " is scanned _ triangle ".

Treatment scheme marches to decision steps 591 to determine whether also having untreated any other limit of sharing as forward sight unit summit CVV.

If in decision steps 591, determine the untreated limit of looking the summit, unit of sharing of residue, then treatment scheme marches to and processes 5-3, it returns treatment scheme to step 575(Fig. 5 A), wherein variable " limit " is configured to refer to the next one of sharing summit CVV and looks element sides.Treatment scheme marches to step 555 to generate next candidate and scan triangle and it is tested.

If determine not have other untreated element sides of looking to share this summit in decision steps 591, then treatment scheme marches to step 592, wherein " current _ polygon " variable set up " is scanned _ triangle " for referring to triangle.

Treatment scheme marches to step 593 and scans triangle with output and " scan _ triangle ".

Treatment scheme marches to step 594 and scans triangular construction SE-MV wedge with basis.The other details of this step is disclosed in Fig. 6 B.

Then treatment scheme marches to processes 5-4, and it is from step 594(Fig. 5 A) begin to advance to next connect look the summit, unit.Then treatment scheme is returned step 535.

If determine to look element sides be not in decision steps 555 from the look silhouette edge of looking from point of MV of point, then treatment scheme marches to and processes 5-2, and it is from step 591(Fig. 5 B) begin to select for the treatment of residue look element sides.

Fig. 5 C shows be used to determining to be formed on inside corner single order profile summit and looking the polygon process flow diagram of the test of support polygon whether between the element sides.

Following alternate embodiments can be arranged: identify the SE-MV support polygon by " side sexual orientation " (with respect to inside of polygonal mesh) of consideration candidate support polygon and the orientation with respect to looking the summit, unit of candidate's support polygon.

In one embodiment, the grid polygon all is assumed to " outside " polygon, it has the normal line vector of " inside " that is oriented partly the district that comprises away from polygonal mesh.In such embodiment, all grid polygons of polygonal mesh as one man have this identical " side " orientation.

As previously mentioned, polygon is a kind of planar structure, and it can have two sides corresponding to the both sides that comprise this polygonal plane.Illustrative embodiments is included as the polygonal mesh of stream shape or sealing.The spatial volume that stream shape grid will be embedded with it is divided into inside and outside.In calculating graphics, it is very useful using with dirty shape grid: the polygonal normal line vector of each in this grid is oriented partly towards the inside away from this closed volume.This can be called as polygonal " outside " side.Opposition side can be called as polygonal " outside " side.If all polygons all have this consistent side sexual orientation in grid, then should can exist never polygonal private side to look as seen from the outside.

In the exemplary embodiment, can namely share the polygonal vertex order on limit and establish the polygon of grid and have identical side sexual orientation by checking adjacent polygon.(referring to Schneider, Philip J., Eberely, David H., " Geometric Tools for Computer Graphics " Morgan Kaufmann2003pp.342-345, its full content merges among the application by reference).If F ₀And F ₁To share by two summit V ₁And V ₂Two adjacent polygons on the limit that forms.For polygon F ₀If, summit V ₁And V ₂With V ₂Follow V ₁Ordering occur, then they are at polygon F ₁In must be with V ₁Follow V ₂Ordering occur.The limit of sharing have this ordering adjacent polygon can be called and be had consistent summit ordering.Polygon with consistent summit ordering has identical side sexual orientation.

In one embodiment, looking the candidate SE-MV support polygon that forms between element sides and the inside corner single order profile summit for single order profile summit.Provide candidate's support polygon side sexual orientation identical with the polygonal side sexual orientation of back side grid of the single order silhouette edge on shared inside corner single order profile summit.(use this consistent side sexual orientation, for example, the people who passes the single order silhouette edge walking on back side grid polygonal " outside " face can run into candidate's support polygon " outside " face).Then check with respect to looking the orientation of summit, unit to the plane of each candidate's support polygon.If the plane of candidate's support polygon is looked the summit, unit with respect to each and just do not faced, the element sides of looking that then forms candidate's support polygon is that element sides is looked in support and this candidate's support polygon is support polygon.

According to some embodiments, enter into the processing shown in Fig. 5 C from step 595.In step 595, on inside corner single order profile summit and look between the element sides (E) and to form candidate's support polygon.Treatment scheme marches to step 596 and is arranged to identical with the polygonal side sexual orientation of back side assembly of the single order silhouette edge of sharing inside corner single order profile summit with the side sexual orientation of candidate's support polygon that will form in step 595.In the exemplary embodiment, the side sexual orientation of SE-MV support polygon can be arranged to consistent with the side sexual orientation of the previous adjacent S V-ME that determines or SE-MV support polygon.Because the polygon that the SE-MV support polygon is adjacent with these is shared the limit, so can be by guaranteeing that adjacent polygon has consistent summit and sorts to arrange the side sexual orientation.

Treatment scheme marches to step 597 to determine whether candidate's support polygon is just faced with respect to each summit of looking in the summit, unit.If determine that in decision steps 597 candidate's support polygons look the summit, unit with respect to each and just do not face, then treatment scheme marches to step 599 and is identified as to support and looks element sides and this candidate's support polygon is identified as support polygon will look element sides (E).

If determine that in decision steps 597 candidate's support polygons just face with respect to looking arbitrarily the summit, unit, then treatment scheme marches to step 598 and is identified as non-support and looks element sides and candidate's support polygon is identified as non-support polygon will look element sides (E).

Fig. 6 A: show the process flow diagram of constructing the method for SV-ME and SE-ME wedge according to SV-ME and the SE-ME support polygon of correspondence

Fig. 6 A shows the process flow diagram according to the processing of the support polygon structure SV-ME wedge of correspondence.This provides the additional detail about the step 116 among Fig. 1.According to some embodiments, from the step 116 of Fig. 1, enter the processing shown in Fig. 6 A.

In some embodiments,, from step 610, wherein the fillet of support polygon is identified as and has for a summit looking the unit and be those limits on another summit, summit of polygonal mesh according to the processing of the SV-ME of correspondence and SE-ME support polygon structure SV-ME and SE-ME wedge.

Treatment scheme marches to step 615 by fillet is begun extend to construct ray from fillet in semi-infinite mode away from looking the unit from the summit of the correspondence of the silhouette edge that supports.If support polygon is triangle, then extend two limits that unit and silhouette edge are looked in connection.If support polygon is quadrilateral (according to the step 470 of Fig. 4 A), then can extend the diagonal line that element sides and silhouette edge are looked in connection.Extend diagonal line and produce larger wedge, in fact it reflected to pass looks the observability that element sides is looked from looking element sides.

Treatment scheme marches to step 620 and is connected to corresponding (support) polygonal mesh silhouette edge to form semi-infinite SV-ME(or SE-ME with the limit of will extend) wedge.

Fig. 6 B: show the process flow diagram of constructing the method for SE-MV wedge according to the SE-MV support polygon of correspondence

Fig. 6 B shows the process flow diagram according to the processing of scanning triangular construction SE-MV wedge of correspondence.According to some embodiments, the step 135 from Fig. 1 enters the processing shown in Fig. 6 B.

In some embodiments,, from step 630, wherein will scan leg-of-mutton fillet and be identified as and have for summit looking the summit, unit and be those limits on another summit on the summit of polygonal mesh according to the processing of scanning triangular construction SE-MV wedge of correspondence.

Treatment scheme marches to step 635 with by these limits are begun extend to construct ray from fillet in semi-infinite mode away from looking the unit from the grid profile summit of correspondence.

Treatment scheme marches to step 640 and forms the semiinfinite wedge with the polygonal mesh inside corner profile summit that the limit that will extend is connected to correspondence.

The structure of following single order wedge has been described in the processing of Fig. 6 A and Fig. 6 B: this single order wedge only is subjected to the restriction of the cross-shaped portion of the adjacent wedge on they and the profile contour.These wedges can be called initial wedge.

According to some embodiments, in subsequent treatment, for example in the structure of single order observability figure, these initial wedges can intersect with grid polygon and other wedges subsequently.Initial wedge can also intersect to form umbra boundary polygon (UBP) with other wedges clearly, and its encirclement comprises (conservatively) the conservative from looking the polyhedron polymerization umbra volume of looking the unit of district that be blocked.

Fig. 7-Figure 11: to the detailed description of the output of the first embodiment

Fig. 7 A shows has summit V ₁To V ₈Protruding unit and the non-convex polygon grid M1 of looking.The silhouette edge that the single order of grid is looked from looking the unit is shown with thick line.Two limits in the single order silhouette edge are labeled as A and B.This is the skeleton view of seeing towards the general direction of polygonal mesh along from looking the unit.

Single order silhouette edge A has with respect at least one and looks the assembly polygon that the summit, unit is just being faced.This assembly polygon is by limit A and is labeled as the triangle that the grid vertex of MV1 forms.By limit A and the triangle that forms at the grid vertex MV2 shown in Fig. 7 B1 for another assembly polygon of limit A.This assembly polygon is with respect to all summit V that look the unit ₁To V ₈The back side pair.Note, these two assembly polygons of sharing limit A are the back side pair relative to each other, so that limit A is the local support limit of polygonal mesh M1 and is the single order silhouette edge.Can be by (for example selecting the first assembly polygon, the triangle that is formed by limit A and summit MV2) and by determine another assembly polygonal be not that the summit (for example, being in this case summit MV1) of sharing the part on limit is positioned at two assembly polygon back sides that the front side that comprises the first polygonal plane or dorsal part determine shared limit A pair.If the summit of not sharing is not positioned at the dorsal part on another polygonal plane of assembly, two assembly polygon back side pair then is as in this case.Can use and make this such as the plane equation of in the definition at " back side " that from vocabulary, provides, describing and determine.In some embodiments, for comprising that each limit among the polygonal mesh M1 repeats processing shown in Figure 3 with each single order silhouette edge of identification polygonal mesh M1.

Fig. 7 B1 shows the polygonal mesh object identical with the described polygonal mesh object of Fig. 7 A M1, but is along the skeleton view from polygonal mesh towards the general direction of looking the unit.According to this figure, limit B has with respect to all summit V that look the unit ₁To V ₈The assembly triangle (being formed by limit B and grid vertex M3) that the back side is right.Shown in Fig. 7 A, limit B has by what limit B and grid vertex MV1 formed and looks another assembly triangle that the summit, unit is just being faced with respect at least one.In addition, these two assembly polygons of sharing limit B are the back side pair relative to each other, so that limit B is the local support limit of polygonal mesh M1 and is the single order silhouette edge.

Fig. 7 B2 shows the polygonal mesh different from the polygonal mesh of drawing in Fig. 7 B1.This polygonal mesh is labeled as M3.The limit of polygonal mesh M3 illustrates and is marked as I with thick line.This limit has for a leg-of-mutton assembly polygon that is marked as T1 with for being marked as leg-of-mutton another assembly polygon of T2.

Assembly polygon T1 just faces with respect to all summits of looking the unit that are labeled as " depending on the unit ", and this is because all looking on the dorsal part that the summit, unit is positioned at the plane that comprises triangle T 1.

Assembly triangle T 2 have on the front side that is positioned at the plane that comprises triangle T 2 at least one look the summit, unit, namely T2 looks the summit, unit with respect at least one and just faces.

Thereby assembly triangle T 1 and T2 satisfy so that the limit that they are shared is with respect to two criterions in the required criterion of the single order silhouette edge of looking the unit.

But because two assembly triangles back side pair not relative to each other, so the limit I that shares is not the single order silhouette edge.This can not be that the summit of sharing the summit on limit is determined by what select triangle T 1 and another assembly triangle (T2) of identification.This summit is P2 in this case.Summit P2 is positioned on the front side on the plane that comprises another assembly triangle T 1.Can use as establishing this fact in the plane equation formula of the triangle T described in the vocabulary description at " back side ".

Because T1 and T2 be the back side pair not relative to each other, so in one embodiment, they are failed in step 345 or 330 by testing in the judgement shown in the exemplary process diagram of Fig. 3.

Fig. 7 C1 shows the figure for the support polygon of single order silhouette edge A and B.Support polygon for single order silhouette edge A is marked as SPA, and is marked as SPB for the support polygon of single order silhouette edge B.Corresponding support is looked summit, unit (SVV) and is labeled as respectively SVVA and SVVB, and it corresponds respectively to looks summit, unit V ₄And V ₈This is the skeleton view of looking towards the general direction of mesh object along from looking the unit.

Fig. 7 C2 is the figure that shows respectively for the support polygon SPA of single order silhouette edge A and B and SPB and corresponding source summit-Grid Edge (SV-ME) wedge.Support polygon for single order silhouette edge A is marked as SPA, and is marked as SPB for the support polygon of single order silhouette edge B.Corresponding support is looked summit, unit (SVV) and is marked as respectively SVVA and SVVB.Be marked as SV-ME WA by the SV-ME wedge that extends support polygon SPA formation.Be marked as SV-ME WB by the SV-ME wedge that extends support polygon SPB formation.According to some embodiments, construct SV-ME wedge WA and WB according to the processing shown in Fig. 1, Fig. 4 and Fig. 6 A.This is the skeleton view of looking towards the general direction of mesh object along from looking the unit.

Fig. 7 C3 only shows the SV-ME wedge that forms by the limit of extending corresponding support polygon.Be marked as SV-ME WA by the SV-ME wedge that extends support polygon SPA formation.Be marked as SV-ME WB by the SV-ME wedge that extends support polygon SPB formation.Corresponding support is looked summit, unit (SVV) and is marked as respectively SVVA and SVVB.This is the skeleton view of looking towards the general direction of mesh object along from looking the unit.

Although Fig. 7 C1-Fig. 7 C3 shows the wedge that is incident on single order silhouette edge A and the B, other embodiment is configured to be included in the wedge of each single order silhouette edge that the single order profile contour among the grid M1 comprises according to the processing shown in Fig. 1, Fig. 3 to Fig. 6 B.

Fig. 7 D1 shows the object identical with object shown in Fig. 7 C1, but is along the skeleton view of looking towards the general direction of looking the unit from mesh object M1.

Fig. 7 D2 shows the object identical with object shown in Fig. 7 C2, but is along the skeleton view of looking towards the general direction of looking the unit from mesh object M1.

Fig. 7 D3 shows the object identical with object shown in Fig. 7 C3, but is along the skeleton view of looking towards the general direction of looking the unit from mesh object M1.

Fig. 7 D4 shows with the polygonal mesh shown in Fig. 7 D3 with identical visual angle and looks the identical polygonal mesh in unit and look the unit.Fig. 7 D4 shows two rotation wedges that intersect at the place, external corners summit of single order profile contour.

One of rotation wedge is marked as SV-ME WA, and it can also be seen in Fig. 7 D3.Other rotation wedge SV-ME WC has been shown in Fig. 7 D4.This wedge is looked the summit, unit by the single order silhouette edge that is labeled as C with the support that is labeled as SVVC and is supported.

Two external corners summits that rotate wedge SV-ME WA and the shared single order silhouette edge of SV-ME WC.This summit is marked as OCV.Such as defined in the step 125 and 140 of the exemplary process diagram of Fig. 1, in one embodiment, the rotating polygon of sharing the external corners summit intersects each other.

The rotating polygon that the summit, unit is looked in shared outer external corners profile summit and rotation to identical support can accurately intersect each other at place, shared limit.In this case, shared limit is to extend from the summit of being shared and be positioned at by support to look the summit, unit and the external corners summit of being shared forms ray on the line.In particular cases this, two rotation wedge each other restrictions on the limit of being shared.

(shared inside corner profile summit and rotation to identical support are looked the rotating polygon on summit, unit and are also accurately located to intersect each other on the limit of being shared.In this case, do not scanning support polygon and do not generating the corresponding wedge that scans.）

In the ordinary course of things, the rotation wedge of sharing the external corners summit can rotate to different supports looks the summit, unit.In Fig. 7 D4, wedge SV-ME WA is by looking summit, unit V ₄Support, and SV-MEWC is supported by SVVC.In this case, the cross-shaped portion of wedge SV-ME WA and wedge SV-ME WC is the line segment that is marked as I.Line segment I is divided into two parts with wedge SV-ME WC.The proximal part of the wedge SV-ME WC that segments is blocked in the figure.

This proximal part of wedge SV-ME WC with different visual angles at Fig. 7 D5 that shows the object identical with the object of Fig. 7 D4 fully as seen.This proximal part is marked as SV-MEWCR in Fig. 7 D5.

Usually, sharing external corners summit and rotation to different support looks the phase intersections of two rotation wedges on summit, unit and causes one of above-mentioned wedge to be constrained to proximal part (for example, SV-ME WCR(indication limit wedge C)) and distal portions.This partly limited wedge only has proximal part to be actually from looking the umbra event face of looking the unit (only having this proximal part is the polygon of corresponding polyhedron polymerization umbra (PAU)).Because this exceeds restriction and all is positioned at the side that is not blocked of adjacent wedge along the distal portions away from the direction of looking the unit, so this distal portions does not represent the umbra event face of looking the unit from looking.In the example shown in Fig. 7 D4 and Fig. 7 D5, in fact the grid polygon that is not blocked side and is blocked on the side that is positioned at the distal portions of SV-MEWC is looked and is not blocked from looking summit, unit SVVA, therefore, looks and is not blocked from looking the unit.

This partial restriction that the adjacent rotation wedge by sharing external corners profile summit of rotation wedge consists of produces fully less wedge in some cases.Because this is less, the limited wedge additional border that comprises that has that restriction processes partly, so it is submitted to carry out the really timing of observability about wedge to work as this wedge, this wedge can need less fully processing (for example, step 1515 place in a kind of embodiment that the 2D trellis traversal shown in the exemplary process diagram of using Figure 15 is processed).

Therefore, this partial restriction is processed and can be accelerated determining about the observability of wedge.Also can adopt the alternate embodiments of not using this partial restriction to process.Any wedge that is not also limited by other wedges still intersects to produce the segmentation of uncontinuity grid with the grid polygon.In fact whether then use polyhedral some pardon of the modification described in the exemplary process diagram of Figure 25 to test to make about this uncontinuity segmentation is from looking determining of the umbra border of looking the unit.This test is fit to local restricted and not limited wedge.

The wedge that the discussion hypothesis of front adopts all is the single order wedge.The high-order wedge stands to intersect (limited by other wedges) such as described wedge-wedge in one embodiment, for example, showing for determining whether actual step 2155 from the exemplary process diagram of looking the method for blocking boundary segmentation of looking the unit of DM_SEG.

Fig. 8 A1 shows the figure that scans triangle (SE-MV support triangle shape) that is positioned on the inside corner summit of being shared by the single order grid silhouette edge that is labeled as A and B of mesh object M1.This scans triangle and is marked as ST_AB.In some embodiments, process to generate with scanning shown in Fig. 5 A and Fig. 5 B and scan triangle ST_AB, scan to be anchored at and be marked as ICSV and from SVVA(V ₄) to SVVB(V ₈) occur.In this case, inside corner grid profile summit be single order profile contour (that is the contour that, is formed by all single order silhouette edges of mesh object M1) share the simple inside corner that the place, summit forms at two single order silhouette edges.This is along from looking the unit towards the skeleton view of the general direction of mesh object, and it is similar with the figure shown in Fig. 7 C1 to Fig. 7 A.

Fig. 8 A2 shows the figure that scans triangle (SE-MV support polygon) on the inside corner summit of being shared by the single order silhouette edge that is marked as A and B of mesh object M1.Scan triangle and be marked as ST_AB.In some embodiments, use to process to generate scanning shown in Fig. 5 A and Fig. 5 B and scan triangle ST_AB, this scans and is anchored at the place, inside corner profile summit that is labeled as ICSV and from SVVA(V ₄) to SVVB(V ₈) carry out.In this case, inside corner grid profile summit is the simple inside corner that forms at place, the shared summit of two single order silhouette edges of single order profile contour.The SE-MV wedge that scans the correspondence that triangle forms by extension is marked as SE-MV WAB.According to some embodiments, form SE-MV wedge WAB according to the processing shown in Fig. 6 B.To this, pass the inside corner summit and the limit of extending polygon ST_AB to form SE-MV WAB.This is along looking the skeleton view of unit towards the general direction of mesh object, and the figure shown in itself and Fig. 7 A and Fig. 7 C2 is similar.

Fig. 8 A3 shows the figure on the inside corner profile summit that is marked as ICSV.Scanning triangle by extension forms corresponding SE-MV wedge and is marked as SE-MV WAB.This is along from looking the unit towards the skeleton view of the general direction of mesh object, and it is similar with the figure shown in Fig. 7 C3 to Fig. 7 A.

Fig. 8 A4 shows the figure of the single order Conservative Linear umbra event face (CLUES) that is incident on silhouette edge A and the B.As shown in Fig. 8 A4, continuous umbra event face is comprised of two SV-ME wedges (being marked as SV-ME WA and SV-ME WB) and single SE-MV wedge (being labeled as SE_MV WAB) in this case.Corresponding support is looked the summit, unit and is marked as SVVA and SVVB, and inside corner single order profile summit is marked as ICSV.This is the skeleton view of looking towards the general direction of mesh object along looking the unit.As shown in Fig. 8 A4, the CLUES that is comprised of SV-ME WA, SE-MV WAB and SV-ME WB forms and blocks the border, and wherein the side that is not blocked on this border is along the direction of arrow U1, and the side that is blocked is along the direction of arrow O1.

Fig. 8 B1 shows the object identical with the object shown in Fig. 8 A1, but from along the skeleton view of looking towards the general direction of looking the unit from mesh object M1.

Fig. 8 B2 shows the object identical with the object shown in Fig. 8 A2, but from along the skeleton view of looking towards the general direction of looking the unit from mesh object M1.

Fig. 8 B3 shows the object identical with the object shown in Fig. 8 A3, but from along the skeleton view of looking towards the general direction of looking the unit from mesh object M1.

Fig. 8 B4 shows the object identical with the object shown in Fig. 8 A4, but from along the skeleton view of looking towards the general direction of looking the unit from mesh object M1.

Identical with Fig. 8 A4, Fig. 8 C shows the figure of the single order Conservative Linear umbra event face (CLUES) that is incident on silhouette edge A and the B.This continuous umbra event face is comprised of two SV-ME wedges (being labeled as SV-ME WA and SV-ME WB) and single SE-MV wedge (being marked as SE-MV WAB) in this case.This is the skeleton view of looking towards the general direction of mesh object along from looking the unit.

Fig. 9 A shows by by Teller(1992) figure that is incident on the umbra event face on silhouette edge A and the B that constructs of the art methods of the non-penumbra of linearization described.In the prior art method that only is used for the finiteness problem of entrance sequence observability, construct umbra event face according to the plane of support polygon fully.The part that is incident on silhouette edge A and the B of these supporting planes is illustrated and is marked as " wedge plane _ A " and " wedge plane _ B ".Intersect with formation and be incident on continuous visibility events face on silhouette edge A and the B at online L place, these planes.

In the non-penumbra art methods of the linearization of Teller, Teller(1992), by only being incident on the entrance limit and being similar to the visibility events face by the Plane intersects that support polygon is looked point-supported in the source, wherein the source be in the entrance sequence at front entrance.These support polygons are corresponding to SV-ME support polygon (using the nomenclature of present embodiment).The method of Teller does not adopt corresponding SE-MV support polygon in the structure of umbra event face, but adopts these polygonal planes.

By contrast, be to be subjected to the semiinfinite for ray of support polygon to extend the laterally semiinfinite polygon of restriction such as the SV-ME wedge of being constructed by present embodiment.The SV-ME wedge also is limited on the corresponding single order silhouette edge.Teller " wedge " is actually does not have the laterally plane of restriction.The present embodiment of structure " Teller wedge " is to extend in the flat sided straight of adjacent S V-ME wedge at inside corner place to above-mentioned Plane intersects.

In the analysis below, show: compare only one type the method for Plane intersects of support polygon of using Teller, by using the visibility events face according to SV-ME and SE-MV support polygon structure, this method can provide the observability solution of looking from the district more accurately significantly.

Must emphasize, method Teller(1992) only is designed to provide the solution of the limited visibility problem that passes polygon entrance sequence.Because in the method for Teller, support the margin of visibility events face in the limit of entrance, so the silhouette edge of visibility events face is constructed in the method nonrecognition of Teller thereon.Because the method for Teller is not used the intersecting plane method and is constructed visibility events face on the silhouette edge of general polygonal mesh; Below the analysis intersecting plane method (supposing to be applied to the general considerations of the observability of looking from the district in the polyhedron environment) that is equivalent to Teller and rotation and scan the visibility events surface construction this method (in fact it be used for more generally visibility problem) theory relatively.

Fig. 9 B shows the figure of the object identical with the object of Fig. 9 A, but from along mesh object towards the skeleton view of looking the unit.

Fig. 9 C shows the more accurate umbra event face that can be produced by this method by the umbra event face that the art methods that intersects supporting plane produces of comparing with Fig. 9 D.In Fig. 9 C and Fig. 9 D, by the rotation of wedge with scan umbra event face that this method of structure forms and be shown as and be superimposed upon on the umbra event face that is formed by the art methods that intersects supporting plane.According to the skeleton view of Fig. 9 D, along looking towards the general direction of mesh object from looking the unit, can find out compared to existing technology method, this method has produced larger, more accurate umbra volume.Compare the only cross-shaped portion of supporting plane, produced larger conservative umbra volume (therefore being more accurate potential visible collection) by the increase of scanning the SE-MV wedge that triangle (SE-MV support polygon) generates.The art methods that is different from intersecting plane is scanned this method of looking unit profile contour and can be explained the effect that comprises about the observability at inside corner profile summit place of looking on the cell surface.Thereby, for any profile contour with inside corner summit of adjacent supports polygon rotation to the different summits of looking the unit, compare the intersecting plane method, this method can produce more accurate result.

Fig. 9 D also illustrates the umbra event face that produced by this rotation and the method for scanning and intersecting plane method by prior art and produces deviation between the umbra event face and tend to along with apart from the distance increase on the silhouette edge that supports and summit and increase.Thereby for most of inside corner profiles summit, the precision of this method can be far above the precision of the art methods of using intersecting plane.

Fig. 9 D shows the object identical with the object shown in Fig. 9 C, but from along the skeleton view of looking towards the general direction of looking the unit from mesh object.

Page turning e-book (flipbook) view of the wedge that the conservative support polygon of identification and structure are corresponding.

When observing with specific order, the subset of Fig. 7 to Fig. 9 provides the conservative support polygon of identification and the page turning e-book view of constructing the method for corresponding wedge.The below enumerates these orders:

Rotary support Duo Bianxing ﹠amp; Wedge: observe from looking the back, unit generally: 7A, 7C, 7C1,7C2,

Rotary support Duo Bianxing ﹠amp; Wedge: observe from looking the front, unit generally: 7B, 7D, 7D1,7D2,

Scan Zhi Chengduobianxing ﹠amp; Wedge: observe from looking the back, unit generally: 7A, 8A, 8A1,8A2, (show the rotation wedge and scan the 8A3 of the combination of wedge).

Scan Zhi Chengduobianxing ﹠amp; Wedge: observe from the front of looking the unit generally: 7B, 8B, 8B1,8B2, (show the rotation wedge and scan the 8B3 of the combination of wedge).

Figure 10 A shows with the grid polygon shown in Fig. 9 A and Fig. 9 B and looks the identical grid polygon in unit and look the unit, but with along the skeleton view from the general direction of the polygonal beneath of grid.Figure 10 A shows the single order visibility events face identical with the single order visibility events face (wedge) shown in Fig. 9 C.Show particularly the SV-ME WA that is incident on the single order silhouette edge A, be incident on SV-ME WB and SE-MV WAB on the single order silhouette edge B.

Also show two other single order SV-ME wedge W4 and W5.It is V that the summit, unit is looked in the support of wedge W4 and W5 ₃Show the cross-shaped portion of these wedges.Wedge intersects each other and intersects to form umbra boundary polygon (UBP) with other grid polygons.These UBP form single order polymerization umbra (PAU) face.The spatial volume that is surrounded by PAU is looked and is blocked for single order from the unit of looking of correspondence.The UBP corresponding with the cross-shaped portion of wedge is not shown clearly, but it can be inferred from intersection is shown in Figure 10 A.Ignore some wedges that meeting in the wedge forms complete PAU with the inner structure (for example, the cross-shaped portion of wedge W4, W5, SV-ME WA, SE-MVWAB and SV-ME WB) of the part that can see single order PAU.

Figure 10 B is the view of the identical polygonal mesh (M1) shown in Figure 10 A.In Figure 10 B, grid M1 is from observing with similar visual angle, the visual angle of Fig. 8 C, observing generally " top " side that comprises the inside corner Grid Edge of grid M1 with looking the unit.This view is very different with the view of looking the unit with the M1 that provides in Figure 10 A.Note, the same edge of M1 is labeled as E and is positioned at grid M1 " bottom " in this two width of cloth figure.In this two width of cloth figure, gone back mark limit A and limit B.

The side that is blocked of wedge has been shown in Figure 10 A.

The side that is not blocked of corresponding UPB has been shown in Figure 10 B.

Figure 10 B shows by making corresponding wedge and other wedges intersect five UBP that form.

UBP-A by the wedge (SV-ME WA) of correspondence and wedge W5(shown in Figure 10 A) cross-shaped portion form.UBP-A also is subjected to SV-ME WA and restriction at the cross-shaped portion of the wedge W4 shown in Figure 10 A.In Figure 10 B, hide W4 fully, but in Figure 10 B, the cross-shaped portion of W4 and SV-MEWA has been depicted as the limit that is marked as F.Limit F is the limit of UBP-A.Additionally, UBP-A and UBP-AB share common edge (it is according to deriving at the SE-MV WAB shown in Figure 10 A).

UBP-AB forms by SE-MV WAB and wedge W4 and with the cross-shaped portion of the wedge of UBP-D.Owing to the structure that scans of corresponding wedge SE-ME WAB, so the shared common edge of UBP-AB and UBP-A and UBP-B.UBP-AB also is subject to itself and restriction corresponding to the cross-shaped portion of the rotation wedge of UBP-D (being supported by Grid Edge D).

UBP-5 is that (at the W5 shown in Figure 10 A, it has corresponding support and looks summit, unit V by the rotation wedge of correspondence ₃) form with W4 and with cross-shaped portion that SV-ME WA forms.

UPB-D is that (not shown wedge is looked summit, unit V but have to support by the wedge that is incident on single order silhouette edge D ₈) cross-shaped portion of the wedge (not shown wedge) that supports with wedge SV-ME B, SE-MV AB and W4 and by limit E forms.

UBP forms the border for the PAU of M1.See in the view that the UBP of the PAU of not every formation M1 can both provide in Figure 10 B.

Figure 10 B shows the wedge that is subjected to other wedge restrictions fully.It is possible using the embodiment of fully limited wedge (for example, the responsive structure of the output of the PAU in the exemplary process diagram of Figure 26).Additionally, (for example use partly limited wedge, the SV-ME wedge that intersects each other at external corners single order silhouette edge place) embodiment for example can be used in the responsive structure of output of the observability figure shown in the exemplary process diagram of Figure 20 A alternatively, the SV-ME wedge that the responsive structure of this output has used the crossing adjacent SV-ME wedge described in the step 140 that can be subjected to alternatively exemplary process diagram shown in Figure 1 to limit.Additionally, because in fact whether described definite wedge and the polygonal cross-shaped portion of grid are the method for blocking the border (adopting the polyhedral some pardon test of revising) in the local or overall situation restriction of making this and not needing the priori of wedge and other wedges before determining, thus can be in the situation that does not have so local wedge-wedge restriction these wedges of use.

Figure 11 A shows in the situation that the figure of the single order visibility events face (wedge) that the compound-contoured contour generates by this method.In this case, SV-ME wedge (" wedge 1 ") is incident on single order silhouette edge A1 upper (supported by single order silhouette edge A1)." wedge 1 " intersects with the single order silhouette edge that is labeled as B1.As discussing in Fig. 2 A, " wedge 1 " is divided into the side that is blocked (B1O) and the side that is not blocked (B1V) with single order silhouette edge B1.This figure is identical with the figure of Fig. 2 A.

Single order wedge " wedge 1 " is the compound-contoured summit that is labeled as CSV with the cross-shaped portion of single order silhouette edge.The compound-contoured summit is corresponding to the inside corner of compound-contoured contour.Use the term of mutationism, CSV is corresponding to the t summit of result's stream shape.Mutationism comprises some singularity (for example, CSV or T summit) on the convection current shape face (for example, stream shape grid) and the research of all line singularities (for example, single order silhouette edge).

" wedge 2 " is single order visibility events face (SV-ME wedge), and it is supported or be incident on the segmentation B1V by the segmentation B1V that is the visible part of single order silhouette edge B1.

Therefore, " wedge 1 " and " wedge 2 " is two SV-ME wedges that intersect at a CSV place.Because " wedge 1 " and " wedge 2 " are to process (Fig. 4 A and Fig. 6 A) by rotation and the rotation of scanning method and construct by using different supports to look summit, unit (being respectively SVV1 and SVV2), so these two wedges do not engage to form continuous umbra visibility events face on the limit.

This rotation and the processing (Fig. 5 A and Fig. 5 B and Fig. 6 B) of scanning of scanning method are used for the SE-MV wedge (SE-MVWA and SE-MV WB) that structure is bonded into " wedge 1 " and " wedge 2 " in continuous umbra visibility events face.Wedge SE-MV WA forms according to being created on the support SE-MV triangle that CSV, SVV1 and support look between the IVV1 of summit between two parties on the unit profile contour (SVSC).Two limit crossing point CSV of this support triangle shape have extended to form semiinfinite wedge SE-MV WA.Similarly, wedge SE-MV WB scans according to being formed on the support SE-MV(that CSV, SVV2 and support look between the IVV1 of summit between two parties on the unit profile contour (SVSC)) triangle forms.Two limit crossing point CSV of this support triangle shape have extended to form semiinfinite wedge SE-MV WB.

The SE-MV WA common edge place that is connected with SE-MV WB connects.SE-MV WA shares common edge with " wedge 1 ".SE-MV WB shares common edge with " wedge 2 ".The wedge of these four connections becomes the part that is incident on silhouette edge A1 and B1V of continuous single order umbra visibility events face.The view of Figure 11 A shows not being blocked of the side that is blocked (arrow O1) of " wedge 1 " and " wedge 2 " (look from looking the unit, single order as seen) side (arrow U1 and U2).The view of Figure 11 A shows " involved " (look from looking the unit, single order as seen) side of SE-MVWA and SE-MV WB.As shown in Figure 11 A, the cross-shaped portion of wedge " wedge 1 ", SE-MV WA, SE-MV WB and " wedge 2 " has formed continuous event face, and arrow U1 and arrow U2 represent the side that is not blocked of this event face.Figure 11 B is the different views of the same structure shown in Figure 11 A.In Figure 11 B, this view is upwards to look towards the side that is not blocked of be blocked side and " wedge 2 " of " wedge 1 ".Figure 11 B also shows " involved " (look from looking the unit, single order as seen) side of SE-MV WA and SE-MV WB.

This has summed up the description of the first embodiment.In this manual, proposed to be used for generating the processing of single order observability face.Other embodiment has specified the polygon of processing grid and limit to generate the order of single order visibility events face.How other embodiment has accurately described in detail with the visibility events face and has determined the polygon and the polygon fragment that are blocked again.In the detailed description of alternate embodiments below, following trellis traversal algorithm being disclosed: is tending to have the single order wedge structure and definite from looking the observability of looking the unit that effectively interlocks in the vertical observability figure construction algorithm of exporting sensitive property.

Figure 11 C show with at two identical polygonal meshs of two polygonal meshs shown in Fig. 2 B, Figure 11 A and Figure 11 B.Fig. 2 B and Figure 11 C show the high-order rotation wedge that is labeled as " wedge _ height ".This wedge is to construct by the visible back-projection approaches of looking the summit, unit that supports of identification that is used in conjunction with Fig. 2 B and relevant figure discussion.In this case, be VSVV for looking the unit apex marker for the visible support of single order silhouette edge segmentation B1V.

Figure 11 A shows the single order rotation wedge " wedge 2 " that is incident on the B1V.Figure 11 A shows by the single order rotation wedge of locating to intersect at compound-contoured summit (CSV) and the continuous umbra event face that scans wedge and become.

Similarly, Figure 11 C shows continuous umbra event face also by becoming at the crossing high-order wedge in place, compound-contoured summit with the single order wedge.In Figure 11 C, by forming the high-order rotation wedge that is labeled as " wedge _ height " in conjunction with the described method of Fig. 2 B at the visible part (B1V) of single order silhouette edge, because " wedge _ height " by the adjusting on the B1V or high-order rotate to form, so it intersects with the compound-contoured summit that is labeled as CSV as the end points of B1V.

Single order wedge " wedge 1U " also is incident on the CSV.In fact, " wedge 1U " is CSV with the cross-shaped portion of whole single order silhouette edge (being shown B1V+B1O).In this case, by making " wedge 1U " (single order wedge, rotate to SVV1) and " wedge _ height " (high-order rotation wedge, rotate to VSVV) with according to intersecting from the wedge that scans that is labeled as SE-MV WC that support polygon forms that scans that scans structure of SVV1 to VSVV by passing CSV, come the continuous umbra face of formation between " wedge 1U " and " wedge _ height " these two wedges.All these three wedges intersect at the CSV place.

The high-order umbra event face of Figure 11 C is compared with the corresponding single order umbra event face shown in Fig. 2 B, it is apparent that, the high-order event face of Figure 11 C produces this larger shadow zone, therefore produces less visible range.If high-order event face and other grid polygons intersect and be used for determining that as seen polygonal those polygons of grid and/or segmentation look conservatively from looking the unit, if then compare the situation of only using the single order wedge, the result can be more accurate observability figure and corresponding PVS.Under this concrete condition, use the high-order wedge to substitute the complexity that corresponding single order wedge does not even increase result's observability figure, this is because in this case, only uses one to scan (SE-MV) wedge and connect two rotation wedges, substitutes required in the single order situation two and scans wedge.

Figure 12: describe in detail-show with the rotation of wedge and scan the alternate embodiments of the method for constructing conservative first-order linear umbra uncontinuity grid.

Figure 12 shows the rotation of using structure single order wedge and scans the process flow diagram of the method for the conservative linearization umbra uncontinuity grid of method construct.According to some embodiments, processing shown in Figure 12 is wherein identified the single order silhouette edge of all mesh triangles shapes from step 1205.In some embodiments, can identify the single order silhouette edge with the method that Fig. 3 describes in detail.

Treatment scheme marches to step 1210 so that be used in the rotation of describing in detail among Fig. 1 to Fig. 6 and the method for scanning is constructed the initial basic wedge that is incident on the single order silhouette edge.In embodiment, basic wedge is to use rotation and scan method construct those wedges on the single order silhouette edge that runs into.In some embodiments, when initial construction, all wedges all are the initial wedges that is not also further limited by the observability step about wedge.

In the method, wedge is differently constructed and limited to uncontinuity gridding method compared to existing technology.In the uncontinuity gridding method of prior art, the plane wedge be not limited at wedge with in the inferior segment: corresponding look unit support structure (summit or limit) looking be blocked from the grid outline elements (summit or limit) that supported.Therefore, these art methods are calculated accurate linear wedges, and it may not form continuous linear umbra event face, and this is because because crossing with corresponding support polygon and grid polygon and so that do not limit the part of wedge.When only considering plane event face, for example, in incomplete uncontinuity grid method (Heckbert1992), these " gaps " in the linear umbra event face are obvious.These gaps are in fact with the limit between two parties that relates to silhouette edge, intersects with support polygon with to look the high-order visibility events face (often being quadric surface) of Bian-Bian between the element sides-Bian event corresponding.When adopting completely uncontinuity gridding, these gaps are filled by high-order event face.

By contrast, in this method according to the wedge of some embodiments structure, only pass through the grid contour structure that supports and support and look cellular construction and limit: any several picture does not between two parties affect the structure of wedge.

In this method of single order uncontinuity gridding, produced in the umbra border obvious gap by incomplete uncontinuity gridding method (Heckbert1992) by with fillings of getting off: 1) the supported contour structure (limit or summit) by ignoring wedge and support are looked geometric figure between two parties between the cellular construction (that is, ignore and geometric figure that the support polygon of wedge intersects) and are come conservatively to limit wedge and 2 between tectonic epochs at wedge) assist the SE-MV wedge on the conservative plane of wedge and the cross-shaped portion place structure of (conservatively) visible grid silhouette edge.This point is called as compound-contoured summit (CSV).This result does not have the linear umbra border " gap ", continuous, conservative that is produced by the incomplete uncontinuity gridding method that only adopts accurate linear event face.

The treatment scheme row marches to step 1215 from step 1210 and is placed in the tabulation that is called " wedge _ tabulation " with the initial wedge that will construct step 1210.

Treatment scheme marches to step 1220 so that first wedge in " wedge _ tabulation " stands to comprise that step 1225 is to the processing of step 1250.In embodiment, implement " wedge _ tabulation " with data structure such as chain type tabulation or the hash table etc. of any expectation.

Treatment scheme marches to step 1225 to determine the visible cross-shaped portion about this wedge of mesh triangles shape and wedge.This cross-shaped portion of mesh triangles shape and wedge is line segment.Visible segmentation (" visible _ segmentation ") about wedge about visible these segmentations of wedge (or its part).

In the method, in some embodiments, determine the 2D trellis traversal method of conservatively visible segmentation by using responsive 1-stream shape (polyline) traversal of output, determine the visible segmentation about wedge.In Figure 14, Figure 15 and Figure 16 and relevant figure in detail, the method has been described in detail and other places are in this manual also discussed it.During carrying out this visible segmentation about wedge and determining method, to identifying from the crossing certain vertex of the silhouette edge looking the unit and look and wedge at its place's single order.These summits are current wedges and are incident on cross-shaped portion between other wedges on the single order profile.The summit of the type is called as the t summit of compound-contoured summit (CSV) and expression profile contour, constructs subsequently auxiliary conservative connection SE-MV wedge thereon.

Treatment scheme marches to step 1235, each " visible _ segmentation " is stored as the encirclement segmentation of single order umbra uncontinuity grid.These segmentations form this grid conservatively being divided into from looking the look district that is not blocked and from looking the look border polyline in the district that is blocked of unit, unit of umbra uncontinuity grids.

Treatment scheme marches to step 1240, and rotation and the method for scanning are used for one or more SE-MV wedges on the CSVS that structure is incident on wedge, and above-mentioned CSVS is identified during the observability step 1225 about wedge.Limit such as the front, each CSV corresponding to current wedge be supported on current wedge intersect from looking the cross-shaped portion of another wedge on the single order silhouette edge of looking the unit.These wedges intersect at CSV point place.

Be used for to generate the SE-MV wedge that is connected to two assembly wedges that intersect at CSV place scan operation with as in conjunction with Fig. 5 A, Fig. 5 B and the described rotation of Fig. 6 B and scan method a part scan operate identical.Scan and occur in the support corresponding with two assembly wedges of CSV and look between the summit, unit (SVV).In some embodiments, the SVV for each wedge determines when constructing (SV-ME wedge).In other embodiments, the SVV for each wedge is at the observability step 1225(SE-MV wedge about wedge, referring to the step 1553 among Figure 15) during determine.

If two wedges rotations of intersecting at the CSV place are to the identical summit, unit of looking, then two wedges accurately intersect at their place, limit, and do not construct new SE-MV wedge.

If two wedges that intersect at the CSV place are to form by rotation to identical two summits looking element sides, then the rotation on CSV and the result that scans structure are single SE-MV wedges.

If two crossing wedges are SV-ME types, then the SE-MV of this connection is conservatively approximate by looking element sides (connect two supports and look the summit, unit) and corresponding to two quadric surfaces that the SV-ME silhouette edge forms of the crossing wedge of CSV.The single SE-MV wedge of constructing at CSV in this case, conservatively is similar to the quadric surface of the correspondence that is formed by the EEE event.In fact, the SE-MV triangle of constructing can be interpreted as having the quadric surface of the degeneration of unlimited pitch.

If two wedges that intersect at the CSV place are to form by rotation to the summit that belongs to difference and look element sides, the then rotation on CSV and to scan the structure result be limit order of connection SE-ME wedge.

If two crossing wedges are SV-ME types, then these connections SE-MV wedge is conservatively approximate by looking element sides and two the quadric surfaces that other silhouette edges form corresponding with the crossing wedge of CSV.Again, each in the SE-MV wedge can be considered to the quadric surface of the degeneration with unlimited pitch of correspondence.

The treatment scheme row from step 1240 march to step 1250 with will be step 1240 all auxiliary initial wedges of structure add in " wedge _ tabulation ", it means that these wedges are finally understood by step 1225 and processes to find visible fragment about wedge.In subsequent step 1250, any SE-MV wedge of constructing in step 1240 is added in " wedge _ tabulation ".

Treatment scheme marches to decision steps 1255 all wedges to have determined whether treated in " wedge _ tabulation ".If residual wedge in " wedge _ tabulation ", then treatment scheme marches to step 1260 to process the next untreated wedge of selecting at step 1260 place in " wedge _ tabulation ", and treatment scheme is back to step 1225 from step 1260.

If determine that in decision steps 1255 all wedges in " wedge _ tabulation " are processed, then treatment scheme proceeds to step 1265 with the observability of looking from looking the unit by a single point in each district in the test single order uncontinuity grid, determines the observability of a single point in each district.In some embodiments, the observability of looking from the unit of each test point is to determine with the some occlusion method shown in Figure 24 B.Reach the polyhedral some pardon test that relevant figure this test described in detail is based on modification in conjunction with Figure 24 B.Importantly, the identical conservative visibility events face (wedge) that structure is guarded that is used for is adopted in this test.

Treatment scheme marches to step 1270, and wherein single order PVS is the set that is not positioned at the segmentation of the mesh triangles shape of umbra (being blocked) district the inside of conservative single order umbra uncontinuity grid or mesh triangles shape.

The comparison of the output sensitive method of the conservative linearization observability figure structure of the non-output sensitive method of conservative linearization uncontinuity cancellated structure and use 3D and 2D trellis traversal

As in Figure 12, describing in detail, can construct conservative linearization umbra uncontinuity grid with the general art methods that is used for structure uncontinuity grid.In the prior art method, construct wedge at each relevant silhouette edge, even those silhouette edges that (in this application for looking the unit) looked and be blocked fully from the source.Then comprise that each wedge of those wedges that are configured on the silhouette edge that is blocked fully and the mesh triangles shape that all intersect are potentially intersected, and subsequently the visible segmentation about each wedge of mesh triangles shape is determined as aftertreatment.

By contrast, use 3D trellis traversal (Figure 20 A and relevant figure), more effectively be used for determining from looking the output sensitive method of the observability of looking the unit to using 2D trellis traversal (Figure 15 and relevant figure) to construct about the observability of wedge and providing from the method for looking the Conservative Linear umbra observability figure that looks the unit.The method has been utilized the intrinsic connectedness of stream shape grid and has been blocked continuity and solved visibility problem with vertical order.The method is determined the visual profile limit and the wedge structure on the visual profile limit interlocks relatively is independent of the output sensitive property of the degree of depth complexity of model with realization.

Usually, the output sensitive process has mainly assessing the cost that size with respect to its input size by algorithm output decides.Because in showing modeling environment, the size of the visual data collection (output) of looking from any vision area is far smaller than the size of whole model (input) usually, is favourable from the output sensitivity of distinguishing the observability pre-calculation process.

In Table V a, summed up the difference of determining two kinds of methods of the observability of looking from the district with conservative linearization umbra event face, the insensitive method of the output of Figure 12 and export the 2D/3D trellis traversal method (Figure 20 and relevant figure) of sensitivity.

Table V a

The comparison of the non-output sensitive method of CLUDM structure and CLUVM structure output sensitive method

Wherein in table and follow-up equation, use following item:

Polygonal quantity in the M=model

N=looks the quantity on the limit in the unit

The quantity of the single order silhouette edge in the S=environment

S _Shaft=the quantity that is formed on single single order silhouette edge and looks the single order silhouette edge in the axle between the unit

M _VThe quantity of the visible polygon the in=model

S _VThe quantity of the visible single order silhouette edge the in=environment

S _VShaft=the quantity that is formed on single single order silhouette edge and looks the visible single order silhouette edge in the axle between the unit

V _wThe quantity on the summit of the cross-shaped portion between=all polygons and the single wedge

M _w=polygonal the quantity of grid that intersects with wedge

S _Svw=about the quantity on (look or look from the limit from point) visible profile summit of wedge

Seg _VwThe quantity about the visible segmentation of wedge of the cross-shaped portion between=grid polygon and the wedge

The table of front is emphasized: for 2D/3D trellis traversal method, during the vertical traversal of convection current shape visible silhouette edge is identified.Thereby, only construct those wedges that supported by visible silhouette edge segmentation.This causes more exporting responsive enforcement.

The comparison of the uncontinuity gridding method of the linearization umbra event face that the uncontinuity grid method of prior art and this use are conservative

The uncontinuity gridding method of prior art has been discussed in the background technology part of this instructions.Uncontinuity gridding method structure umbra and penumbra visibility events face and determine them with the polygonal visible cross-shaped portion about wedge of grid.The heavy grid division polygons of these cross-shaped portions so that in each face of uncontinuity grid or " district " view (" back projection situation ") in source be equal in topology.The target of the uncontinuity gridding method of prior art is the brightness uncontinuity that mainly identifies in the penumbra region of present area source.

In some embodiments, this observability pre-computation methods of looking from the district is not with penumbra visibility events face but alternatively only to identify from looking the unit conservatively visible grid polygon segmentation of looking with the umbra visibility events face of guarding.Can construct the umbra uncontinuity grid of guarding of describing as among Figure 12 (the uncontinuity cancellated structure of non-output sensitivity) and Figure 19, Figure 20, Figure 21 and relevant figure (output responsive from looking the observability figure structure of looking the unit) with these event faces.Alternately, conservative umbra wedge can intersect each other to form the umbra boundary polygon (UBP) as describing in Figure 26.

Table V b has provided the comparison of Conservative Linear umbra observability figure (shown in Figure 20 and relevant figure) with the uncontinuity gridding method of prior art.

The row that is labeled as " wedge of generation " illustrates: use 2D trellis traversal (Figure 15 and relevant figure) and use this method of 3D trellis traversal (Figure 20 and relevant figure) all to comprise the observability method of looking from the district, because the observability face only is created on visible (not being blocked) (single order) silhouette edge, above-mentioned observability method of looking from the district is relatively to export sensitivity.The uncontinuity grid method that the event face of this and prior art is formed on all (general look from the district) silhouette edges forms contrast.

Table V b

Incomplete and the completely contrast of uncontinuity gridding method of Conservative Linear umbra observability figure (CLUVM) and prior art

In order to sum up, this single order uncontinuity gridding method has been eliminated the gap of finding in the umbra border that the incomplete uncontinuity gridding method of only considering linear summit-Bian visibility events face by prior art produces.The single order method is by filling these gaps to get off: 1) by ignoring the visible part and 2 that conservatively extends wedge about blocking of wedge in the corresponding support polygon) at the wedge planar S E-MV wedge conservative with the cross-shaped portion place structure of (conservatively) visible grid silhouette edge.These cross-shaped portions of wedge are CSV, and the SE-MV wedge structure that uses the method for scanning to carry out at these aspects has produced the continuous linear umbra border in the corresponding uncontinuity grid.Except be similar to and always the conservative interstitial area, this border is accurate.

The gap of the appearance in the incomplete uncontinuity gridding method of prior art in the umbra border is corresponding to the secondary umbra event face that provides in uncontinuity gridding method completely.But quadric use has increased the complexity of implementing fully.

Be different from the linear event face that in the incomplete uncontinuity gridding method of prior art, uses, the conservative linearization umbra event face that adopts in the method for Figure 12 forms continuous umbra face, its make it possible to be blocked and conservatively visible polygonal mesh segmentation determine.

As illustrating in this manual subsequently, embodiment also comprises the precision of the conservative linearization event face in these interstitial areas of adaptively refinement.Use this refinement and these plane wedges only, the method can be constructed in interstitial area and be converged on accurate secondary event face with the conservative umbra event face of the error margin that is positioned at appointment.

The quadric completely uncontinuity of needs compared to existing technology grid method uses conservative linearization umbra event face to simplify greatly any enforcement.But interested is to notice, although simplified enforcement, when from the actual quantity aspect of required event face, the progressive complexity of linearization umbra face with only use the complexity of the quadric completely continuity of S-EEE grid method suitable.

Table VI a shows the upper bound by the quantity of the umbra event face of single order method generation and uncontinuity net boundary, but the upper bound of the quantity of the S-EEE face (quadric surface) that its participant generates in the continuity grid is implemented is completely compared, and the hypothesis quadric surface only is created on the single order silhouette edge in this continuity grid is implemented.

Polygonal quantity in the M=model

N=looks the quantity on the limit in the unit

The quantity of the single order silhouette edge in the S=environment

S _Shaft=the quantity that is formed on single single order silhouette edge and looks the single order silhouette edge in the axle between the unit

Table VI a

The various observability propagation models that use the responsive structure of the non-output of hypothesis are to the estimation for the quantity in the event face of uncontinuity grid and unit/district

Last two row of table show the event face that produces and the expection complexity of uncontinuity net boundary when the plane with adjacent S V-ME wedge of locating to adopt Teller on compound single order profile summit (CSV) extends to the linearization technique of intersection.Be different from single order rotation and scan method and adopt quadric S-EEE method, the method that the SV-ME wedge is extended to the intersection can at the new event face of CSV place introducing, therefore, can produce less event face.The precision of the event face of these extensions generally precision than corresponding single order SE-MV wedge (especially near looking the unit) is low.Thereby, if then they can tend in fact to intersect with relatively large model polygon and employed this shadow zone near looking the unit then underestimate significantly this shadow zone.In the present invention who adopts following technology, use this fact: with the close single order SE-MV(at CSV place by rotating and scanning to generate) wedge.From looking the unit at a distance, the SV-ME wedge that intersects at the CSV place extends to the intersection.Result's extension wedge is not having such out of true apart from the larger distance of looking the unit.

The third line of Table VI a illustrates the completely uncontinuity grid that comprises non-S-EEE event face and has considerably higher complexity.

The quantity of the conservative estimation hypothesis silhouette edge of the quantity of the event face of propagating for the single order observability shown in the Table VI a is the linear function of the polygonal quantity in the model.In fact, first order modeling only generates the event face on the single order silhouette edge, compares the general silhouette edge of looking from looking the unit by completely uncontinuity gridding method employing, and it is not a lot of significantly.For the grid that describes in detail, the quantity of silhouette edge is easy to the square root near polygon quantity.If make this hypothesis, then the quantity of event face can reduce by an order of magnitude, and the complexity that the uncontinuity grid is arranged can reduce by two orders of magnitude.This hypothesis is the basis of estimation of the complexity of the given single order complexity map of Table II.

Certainly, also facilitate uncontinuity grid completely complexity be the following fact: in (non-output is responsive) building method of prior art, construct all visibility events faces, even the many visibility events faces in these visibility events faces virtually completely be positioned at look from the district (for example, look from looking the unit) umbra, therefore do not facilitate the actual umbra of looking from the district or penumbra border.By contrast, the output sensitive method of structure CLUVM generates single order umbra event face, its never other (single order) visible umbra event faces to be comprised on the meaning among the CLUES be extreme value, although CLUES can come refinement to explain high-order observability effect by back projection.

Again, the responsive structure of non-output of Table VI a hypothesis uncontinuity grid.Other places in the present note, the responsive structure of output that corresponding observability figure is shown can reduce the quantity of the event face of constructing and the complexity that corresponding district arranges significantly.

The observability method of looking from the district in some embodiments, is only calculated the conservative approximate of accurate umbra border incident face or these faces.By contrast, the uncontinuity grid method must generate all event faces (umbra and penumbra), and itself and polygonal mesh intersect so that this grid heavily is divided into the uncontinuity grid surface, and the topological view in source region is unified in above-mentioned uncontinuity grid surface.Thereby (accurate or conservative) umbra boundary surface generally is the little subset of the visibility events face that calculates in the uncontinuity gridding method.

By structure umbra uncontinuity grid, can come with the polygon direction inherently only construct umbra event face (single order or accurately) in the situation that do not need to contain functional test from looking this method that the observability figure that looks the unit or polyhedron polymerization umbra calculate the observability of looking from the district.In some embodiments, these umbra event faces are the little subsets that limit the incomplete collection of visibility events face.Thereby, comparing uncontinuity grid completely and carry out aftertreatment with the method on identification umbra border, this method is to calculate the more effective method of the observability of looking from the district.

Figure 13 shows the processing of identifying and differentiating overlapping circulation during the 3D trellis traversal.

Adopt the output sensitive method (shown in Figure 19 to Figure 21) from looking the observability figure structure of looking the unit of 3D trellis traversal can run into the overlapping circulation of polygon (for example, in the step 1935 of Figure 19 and in the step 2015 of Figure 20 A).The overlapping circulation of such polygon can also run into the Weiler-Atherton algorithm of determining method for the observability of looking from point of prior art.

Because be partial to towards jumping to element nearer, that block potentially, so it is subject to be blocked by the cyclic part of grid elements the impact of the infinite loop that causes in the trellis traversal method shown in Figure 19 to Figure 21.For example, consider following situation: potential thing and the potential shelter grid elements of being blocked both is restricted to the polygon cluster that is included in the bounding box.Be apparent that, three (3D) bounding boxs can be arranged in the three dimensions so that these boxes partly block each other in the mode of circulation.For example, consider three boxes, box A is partly blocked by box B.Box B is partly blocked by the 3rd box C.If box C is partly blocked by box A, then three boxes form one and block circulation.If trellis traversal is processed in the axle test and adopted these bounding box elements, then between the step 2005,2010 and 2030 of Figure 20 A, can produce infinite loop.

In embodiment, can adopt two kinds of technology to eliminate by the circulation of grid elements and block the infinite loop that causes.In one approach, reduce to be used to the size of the element of axle test.This can realize with the bounding volume hierarchy (BVH) of sub-box naturally by lower in hierarchical structure.Less grid elements unlikely produces and blocks circulation.For management cycle, come dynamically element to be carried out refinement from larger cluster to less cluster even to each triangle.In some embodiments because even three independent triangles just can cause blocking circulation, the circulation that all these technology are not eliminated under all situations is blocked.

Under these circumstances, the second technology has effectively been eliminated and has been blocked circulation, in this technology, grid elements is defined as triangle, and the overlapping relationship between the triangle is maintained in digraph.To near leg-of-mutton any redirect, check for the relative section of circulation to this digraph.This inspection can be adopted the algorithm of Tarjan or be used for identifying at digraph other O (n) time algorithm of circulation (the strong assembly that connects).In case identified the overlapping circulation of triangle, can identify to interrupt this circulation by all triangles to the triangle of the violation triangle (the root triangle of circulation) that causes circulation-look in the unit shaft.All triangles in this axle all intersect with this axle.This cross-shaped portion is subdivided into the assembly that is positioned at axle fully with these triangles and is positioned at the outer assembly of axle fully.Then with can not again initiating traversal with the triangle of these segmentations of violating triangle formation circulation.This is because must be positioned at outside axle or the axle with any triangle of violating triangle formation circulation.Can also adopt this technology to differentiate the circulation of blocking on grid elements rather than each triangle.

In another kind of technology, in order to eliminate circulation, often directly adopt the leg-of-mutton rear a kind of method in other leg-of-mutton axles of finishing.The needs that this has eliminated structure and has safeguarded the digraph of storage overlapping relationship.But it can produce nonessential cross-shaped portion/finishing and calculate, for example, and when there not being circulation time.

Now turn to Figure 13, Figure 13 and Figure 20 category-A seemingly, for the major part of the 3D trellis traversal algorithm of the responsive structure of the output of the observability figure that looks from the district.Figure 13 detects and differentiates so overlapping step and come Figure 20 is replenished by the step that may run into the overlapping circulation of polygon and the interpolation that shows the processing of 3D trellis traversal.

In some embodiments, processing shown in Figure 13 from step 1305, the mesh triangles shape that wherein in the main processing of 3D trellis traversal (step is identical with step 2010 among Figure 20 A), runs into.

Treatment scheme marches to decision steps 1310 to determine by current triangle with look any whether the triangle that does not travel through is arranged in the axle that the unit forms.This step is identical with step 2010 among Figure 20 A.

If in decision steps 1310, determine have the not triangle of traversal in the axle that unit and current triangle form by looking, then treatment scheme march to step 1330 with on cover in the digraph that triangle is inserted into the overlapping relationship between the expression triangle.Carry out step 1330 and follow-up step and the corresponding step 2030 among Figure 20.

At step 1330 place, treatment scheme marches to step 1335 to determine whether triangle forms circulation in the digraph of expression overlapping relationship, if so, then treatment scheme marches to step 1340.In some embodiments, can use the prior art algorithm in linear session, to carry out for the circulative inspection in the digraph.The appearance that circulates among this figure represent to relate in current triangle and the axle on to cover leg-of-mutton circulation overlapping.

In step 1340, on cover triangle and be trimmed to two assemblies: be positioned at an assembly of axle fully, and be positioned at another outer assembly of axle fully.This has effectively eliminated to design and has covered leg-of-mutton any potential overlapping circulation on original.

Follow step 1340, treatment scheme marches to step 1345, and wherein the 3D trellis traversal jumps to the assembly triangle that covers.This is identical with step 2030 among Figure 20 A.

Do not form circulation with current triangle if in decision steps 1335, determine to cover triangle.Then treatment scheme directly marches to step 1345.

If determine that in decision steps 1310 then treatment scheme marches to step 1315 at triangle-the look triangle of traversal not in the unit shaft.Step 1315 is identical with the corresponding step of Figure 20 A with follow-up step.

As previously mentioned, with polygon-look unit shaft to identify circulation.Must carefully construct polygon and look the axle between the part that comprises on the rear side on this polygonal plane of unit.In some embodiments, divide the polygon that this looks the plane of unit equally for being arranged in, only use the part on the rear side that is positioned at this plane of looking the unit.Identify in this way polygon in this method-look unit shaft.

If in decision steps 1310, determine by current triangle with look and do not have the not triangle of traversal in the axle that the unit forms, then needn't check and circulate overlapping and treatment scheme marches to step 1315.When using traversal and construct the observability figure that looks from the district shown in Figure 20 A, decision steps 1315 is identical with decision steps 2015 among Figure 20 A.In this case, the step 1318 that is labeled as " processing cross-shaped portion " is corresponding with the step 2035 of Figure 20 A, wherein travels through blocking the boundary termination.Step 1320 among Figure 13 is identical with the step 2020 of Figure 20 A.The step 1322 that is labeled as " processing limit " corresponding to the step 2045 among Figure 20 A to 2070.

Weiler-Atherton algorithm and this 3D trellis traversal method (being used for structure observability figure, uncontinuity grid or PAU) both adopt for the overlapping inspection of circulation in some embodiments.

The Weiler-Atherton algorithm is image-space, the object precision methods of determining from the observability that point is looked, and it has general degree of depth complexity assessing the cost for (except the convex region of grid) non-output sensitivity with respect to scene.

When being used in combination, this method of 3D trellis traversal (Figure 20 and relevant figure) is the computing method of looking from the district with 2D trellis traversal (Figure 15 and relevant figure), it is generally responsive for output with respect to the degree of depth complexity of observability figure.(can also 3D/2D trellis traversal method be applied to the visibility problem of looking from point by simply viewpoint being used as the unit of looking of degenerating).

Table VI b shows the Weiler-Atherton algorithm and uses some differences between this method of 3D/2D stream shape trellis traversal.The first row of this table emphasizes that the Weiler-Atherton algorithm generates visibility events face (clipping edge) effectively at each place, polygonal every limit, except ignoring the protruding shelter of non-silhouette edge.By contrast, 3D/2D stream shape trellis traversal method is utilized the intrinsic continuity of blocking of stream shape by only generate the visibility events face at the silhouette edge place.

In addition, by carrying out vertical depth order and the traversal by staggered visible polygon and blocking the structure on border, 3D/2D trellis traversal method has realized comparing the performance specification that the Weiler-Atherton algorithm export sensitivity more.

Execution is the cost of testing when the polygon of front shape blocking potentially for the cost of the strict vertical traversal of 3D/2D trellis traversal algorithm.This in fact class axle test, this axle is formed on current polygon and looks between the unit.Can accelerate such axle test with the hierarchical space segmentation or with the encirclement volume that mode like the technology type that is used for the rejecting of acceleration view frustums is carried out.Use the method, the cost tendency of axle test and the O (logN) of the quantity N of the element that blocks potentially.NlogN for the N that will test a current polygonal overall cost.This strict vertical order has been guaranteed the current polygonal quantity N that will test near the quantity of visible polygon, has therefore produced output susceptibility.

Table VI b

Trellis traversal is implemented: 2D trellis traversal (Figure 14, Figure 15 and Figure 16)

With reference to Figure 14 and relevant Figure 15 and Figure 16, the 2D trellis traversal processing (traversal of 1-stream shape) that is used for constructing about the visible 2D observability of wedge figure is described.

The 3D trellis traversal is processed (describing in conjunction with Figure 19, Figure 20 and relevant figure in this manual) and has been solved the visibility problem that 3D looks from looking the unit.In some embodiments, the output that the 3D trellis traversal is processed is by the conservative linearization observability figure that forms from looking the polygon segmentation that is not blocked of looking the unit, and above-mentioned is by whether crossing the and definite wedge cross-shaped portion reality of conservative linearization umbra event face (CLUES) or wedge and grid polygon is identified from looking the border of blocking of looking the unit from looking the polygon segmentation that is not blocked of looking the unit.

As described in some embodiments, CLUES can also intersect with the grid polygon and need to determine that this cross-shaped portion is whether from looking the border of blocking that the unit looks.The polygonal division to grid of result in these embodiments is a class uncontinuity grid, subsequently it is carried out aftertreatment to determine as seen whether grid elements look from looking the unit.By comparing, the 2D trellis traversal has dealt with the 2D visibility problem: about the visibility problem of wedge, it is similar with the visibility problem about wedge of finding in the uncontinuity gridding method of prior art.In the method for prior art, be to solve in following simple and different mode about the visibility problem of wedge: triangle and wedge are intersected, then use 2D Weiler-Atheron algorithm (it is not that output is responsive) to determine visible segmentation.By utilizing the intrinsic observability correlativity of 1-stream shape, this 2D trellis traversal method is with problem more effective, that the responsive mode of output has solved same light.

Be equivalent to according to element (depending on summit, unit or limit or inside corner profile summit) and pass the visible segmentation about wedge that polygonal mesh element (limit or summit) is sought about the observability of wedge.In some embodiments, wedge is the linearization umbra event face (CLUES) of guarding.

If wedge is by looking the summit, unit and the grid silhouette edge forms, then wedge being called the SV-ME wedge.The SV-ME wedge has proposed the visibility problem about wedge of looking from point.If wedge is by looking element sides and grid vertex forms, then wedge being called the SE-MV wedge.In some embodiments, the wedge of the type has also proposed the visibility problem about wedge of looking from point, and this is because this observability is that the viewpoint of looking from the grid silhouette edge is along solving away from the direction of looking the unit.

The 3rd class wedge (SE-ME) is to look in particular cases by this two limits formed parallel with the grid silhouette edge of element sides.In this case, the rotation of structure umbra wedge is processed and is not run into unique support and look the summit, unit but alternatively run into and look element sides.The such wedge of nonrecognition in the uncontinuity gridding method of many prior aries, if aim at world's axis by chance but look the unit, then such wedge can distinguishingly be present in the architectural environment (wherein many limits tend to parallel or perpendicular to ground level) certainly.

Difference and its two classes wedge, SE-ME wedge have proposed following visibility problem about wedge of having looked from the district: look the observability about wedge that element sides is looked from support.In some embodiments, by to be in general position look the unit is heavily divided or by the SV-ME wedge that the SE-ME wedge is considered as distinguishingly connect and has a lexicographic ordering to eliminating the SE-ME wedge.

In the method, SE-ME wedge and be directly to process by 2D stream shape traversal method about the visibility problem of looking from line segment of wedge, it has solved the observability (for SV-ME and SE-MV wedge) of looking from point and the observability (SE-ME wedge) of looking from line segment.

Compare the situation of looking from point, the situation of looking from line segment is inherently more loading, even this is because in the observability of 2D(about wedge) in the visibility problem of looking from the district also can relate to the high-order visibility events face of advising such as people such as Koltun.

This 2D stream shape traversal method has solved the visibility problem about wedge for the situation of looking from point and the situation of looking from the limit.For the situation of looking from the limit (SE-ME wedge), in some embodiments, the method only adopts the single order umbra event face that is called as wedge line (WL), or is called as the high-order event face of umbra boundary line (UBL).Figure 14, Figure 15 show relevant process flow diagram with Figure 16, it comprises for solving for the wedge of looking from point (SV-ME wedge, wherein supporting and looking unity element is the summit) and both 2D stream shape traversal methods about the visibility problem of wedge of the situation of the wedge (SE-ME wedge, wherein supporting and looking unity element is the limit) of looking from the limit.Except in fact " element " that solves visibility problem be not to look the summit, unit but the inside corner grid profile summit, the SE-MV situation is identical with the SV-ME situation of looking from point.

Figure 17 and Figure 18 show the process flow diagram of following methods: " adjusting " single order WL increases precision about the observability solution of wedge with approximate high-order UBL with back projection conservatively.Propose in the details of Figure 17 and Figure 18 further part in this manual.

As to be applied to the output of determining about the visible mesh triangles shape of wedge cross-shaped portion 2D traversal processing be the 2D observability figure that guards, and it is by being 1 visibility events structure about the visible ambiguity line segment of wedge and the conservative degree of freedom that is called wedge line.Polyline is the 1-stream shape of traversal.The ambiguity line segment is corresponding with the visible cross-shaped portion on wedge of wedge with mesh triangles shape.In 2D stream shape traversal method, can create polyline polymerization umbra (PLAU) by WL with each 1-stream shape and other 1-stream shape " fusion ", realized that therefore the shelter on the plane merges.

Process flow diagram shown in Figure 14 shows shown in Figure 15 and has carried out the initiation of the main 2D traversal processing of detailed description, the processing of having retransmitted and having stopped at the further part of this instructions.

According to some embodiments, processing shown in Figure 14 wherein shows selection " seed " segmentation and initiates to process for the 2D trellis traversal of single wedge from step 1405.

In a kind of embodiment of 2D traversal method, traversal start from by also be connected to (adjacent to) the visible segmentation that forms of the wedge of the current wedge the processed mesh triangles shape of intersecting.In the method, will be used for as the mesh triangles shape of the part of the polyline polymerization umbra (PLAU) of adjacent wedge initiating 2D traversal to current wedge.Because this grid polygon with for apart from for adjacent wedge look the nearest crossing visible segmentation in unit, so this grid polygon is likely for the nearest segmentation of current wedge.Use known adjacent, nearest crossing element to initiate traversal and promoted the performance of processing.The segmentation that is connected to this initial fragment forms 2D kind sub-grid and at an easy rate by determining that the leg-of-mutton cross-shaped portion adjacent with the seed triangle that also intersects with this wedge identify.

Treatment scheme marches to step 1410 to identify comprising a little of 2D kind sub-grid by the cross-shaped portion of wedge horizontal boundary and 2D kind sub-grid.

During comprising on having identified seed 2D grid and this kind sub-grid, treatment scheme marches to step 1415 and processes to call the main of 2D trellis traversal.Figure 15 illustrates main 2D trellis traversal.In some embodiments, use seed 2D mesh triangles shape/segmentation of selecting in the step 1405 of Figure 14 to initiate main 2D stream shape traversal processing shown in Figure 15.In addition, can end at the particular fragments place of 2D grid main 2D stream shape traversal processing (at step 1530 place) shown in Figure 15.Processing shown in Figure 14 determines whether remain any suspended traversal (step 1420) and whether should again initiate main 2D stream shape traversal processing (step 1430 and step subsequently) in these suspended segmentations in main 2D stream shape traversal processing.

If determine not deposit the segmentation of the major part of having ended the 2D trellis traversal thereon in decision steps 1420, then treatment scheme marches to the step 1425 that stops for the 2D trellis traversal of this wedge, thereby has finished the structure of 2D about the observability figure of wedge.

If determine to deposit the 2D grid segmentation that main 2D trellis traversal has thereon been ended (in the step 1530 of Figure 15) in decision steps 1420, then treatment scheme marches to decision steps 1430 to determine looking in the axle that unity element (VCE) forms whether have any triangle by the segmentation (SEG) of ending with as what look summit, unit (SV-ME wedge) or look element sides (SE-ME wedge).Alternately, if this wedge is the SE-MV type, then for the purpose of this algorithm, in fact solve according to the inside corner grid profile summit that is regarded as " VCE(point) ".

If determining in decision steps 1430 does not have triangle in the axle between SEG and VCE, then treatment scheme marches to step 1455 with comprising a little on the 2D grid (polyline) that forms in the segmentation that is connected to current termination segmentation.

Treatment scheme marches to step 1460, wherein again initiates the 2D trellis traversal in formerly the termination segmentation.

If have triangle in decision steps 1430 is determined axle between SEG and VCE, then treatment scheme marches to step 1435 to determine whether any not triangle/segmentation of traversal is arranged in the SEG-VCE axle.

If determine in the SEG-VCE axle to exist the not triangle/segmentation of traversal in decision steps 1435, then treatment scheme marches to step 1440, wherein " redirect " or restart the 2D trellis traversal in the triangle/segmentation of the nearer not traversal in the SEG-VCE axle.This step (it also appears in the step 1530 of main 2D traversal processing shown in Figure 15) guaranteed continue before the traversal and making the decision that stops traversal (because end segmentation look from VCE blocked fully) before treated all triangle/segmentations of blocking potentially.

If in decision steps 1435, determine in the SEG-VEC axle, not exist the not triangle of traversal, then treatment scheme marches to step 1445 determining whether current (terminations) segmentation is positioned at any polyline polymerization umbra (PLAU), its represent wedge from the VCE(of wedge or the inside corner profile summit in SE-MV wedge situation) one or more polygons subarea of looking and being blocked.In some embodiments, the 2D version of polyhedral some pardon test of modification shown in Figure 25 is used in this test.The wedge line (WL) of single order event face (single order wedge) or the correspondence in the 2D situation is used in polyhedral some pardon test of this modification; Alternatively, the umbra boundary line (UBL) of high-order umbra polygon (UBP) or the correspondence in the 2D situation.This test allows needn't construct whole polyhedron/polygon to the implicit solution of the some pardon problem of multiaspect or polygonal some pardon problem.

If determine that in decision steps 1445 point of ending in the segmentation is positioned at PLAU, then treatment scheme marches to step 1450, does not wherein again initiate traversal and no longer will end segmentation markers for ending in the termination segmentation.

Be not positioned at any PLAU if determine to end segmentation in decision steps 1445, then treatment scheme marches to step 1455, and wherein the 2D grid that is associated of structure comprises a little and again initiate as previously mentioned and travel through.

The 2D trellis traversal is implemented: main process (Figure 15) of 2D traversal

As previously mentioned, as shown in figure 14 processing is used for being shown specifically and being called as initiation that main 2D trellis traversal processes at Figure 15 in step 1415, again initiating and termination is controlled.

In some embodiments, processing shown in Figure 15 from step 1505, the traversal of the segmentation that wherein starts from not traveling through about 2D grid (polyline) traversal of the mesh triangles shape segmentation of wedge.Forward from the polyline traversal that is fragmented into segmentation, along advancing from one of both direction of being fragmented into adjacent sectional.The 3D traversal (having a detailed description in the discussion of Figure 20) of the correspondence of triangular mesh can be passed three adjacent boundaries and be occured, and for example fills in (flood-fill) or the breadth first traversal at flood and advance.

Treatment scheme marches to decision steps 1510 to determine in current segmentation and to support the triangle/segmentation that whether has any traversal (processing by current 2D trellis traversal) in the 2D axle look between the unity element (segmentation-look unity element axle or SEG-VCE axle).Single initial wedge and the cross-shaped portion of a plurality of visible mesh triangles shapes can produce a plurality of ambiguity line segments and a plurality of corresponding 2D axles potentially.In a kind of enforcement, construct 2D axle (SEG-VCE axle) for each segmentation in these segmentations.Following alternate embodiments can be arranged: the single conservative axle that will surround the segmentation of a plurality of connections is used for the axle of accelerating step 1510-comprise test.If have any not triangle/segmentation of traversal in the axle in larger comprising, then process each axle that can be back to for each segmentation.

In some embodiments, by adopting hierarchical space segmentation (and/or bounding volume hierarchy (BVH) structure) tissue 3D polygonal mesh geometric figure to comprise test with the axle of further accelerating step 1510.Any hierarchical organization in usage data storehouse is easy to axle is comprised that the cost of test is reduced to about NLog (M) from about N*M, and wherein M is that polygon quantity in the model and N are the quantity of the axle tested.

According to some embodiments, if initial wedge is the SV-ME wedge, then corresponding SEG-VCE axle according to segmentation with support the Support Level formation of looking between the summit, unit.In other embodiment, if initial wedge is the SE-MV wedge, then corresponding SEG-VCE axle forms according to segmentation and Support Level between the grid profile summit of supporting, and this is because of in this case, and observability is from looking element sides and passing grid profile summit and look.If original wedge is to be created on the silhouette edge that supports to look the parallel SE-ME type in particular cases of element sides with support, then the SEG-VCE axle is according to segmentation and supports the Support Level formation of looking between the element sides.

If there is the not segmentation of traversal in the SEG-VCE axle, then the treatment scheme row proceeds to step 1530, and wherein the 2D traversal jumps to segmentation nearer in the SEG-SILE axle.These two steps are carried out vertical processing order.As will be described, this is similar with step 2010 and the step 2030 processed for the 3D trellis traversal.Note the similarity between Figure 15 and Figure 20.Although 3D traversal use 3D mesh triangles shape-look unit (TRI-VC) axle to test nearer not traversal element, this 2D traversal adopts 2D SEG-VCE axle to test nearer not traversal element.

If determine not exist not the segmentation of traversal in the SEG-VCE axle in decision steps 1510, then whether treatment scheme is advanced and is proceeded to step 1515 and comprise with the segmentation of determining current traversal and block border (OB) and put or comprise frontier point.

The containing frontier point appears at the intersection on the border of ambiguity line segment and wedge.

OB point expression 2D ambiguity linear flow shape is upper looks visible and from look border between being blocked of VCE from VCE.The OB point can appear at the intersection of the polygonal mesh segmentation on WL and the wedge.WL and be corresponding 2D(about wedge with each cross-shaped portion of polygonal mesh segmentation) the 1D element (point) of uncontinuity grid.The same with any visibility events face, WL represents the observability of looking from specific source feature (summit in the 2D situation).According to some embodiments, if VCE looks element sides in SE-ME wedge situation, then in the initial point place limit of the WL on the profile summit of correspondence, WL not necessarily represents on the wedge to look visible and from looking look border between the district that is blocked of element sides from looking element sides.

This is because for any visibility problem of looking from the district, and the substantial portion that look from the district (district of element sides is looked in conduct in the 2D of SE-ME wedge situation) in space is blocked is not only by event face (WL) and the polygonal cross-shaped portion of grid and is determined and determine by the cross-shaped portion of WL and other WL.These cross-shaped portions have formed together and to have surrounded from looking the 1-stream shape face in this shadow zone that element sides looks.These 1-stream shape faces are called polyline polymerization umbra (PLAU).Form PLAU according to the segmentation that is called as umbra boundary line (UBL) that can be constructed by the WL of correspondence.

Thereby not every 2D uncontinuity net point (2DDM) point (being positioned at WL and intersection about the triangle segmentation of wedge) all is to block border (OB) point.In subsequently with the step 1563 that discusses in detail, make the identification that the OB of reality is ordered.

Comprise the OB point if determine current segmentation in decision steps 1515, then treatment scheme marches to step 1535 to interrupt the current 2D trellis traversal at OB point place.In other embodiment, treatment scheme marches to from step 1535 and processes 15-1, and it is back to step 1525 with treatment scheme.Similarly, at decision steps 1515 places, whether the current segmentation of determining the 2D grid comprises comprises the border.This frontier point is the cross-shaped portion of the boundary line of 2D grid and wedge, such as the processing of the main 2D traversal processing of the step 1410 in Figure 14 or 1455(control Figure 15) in determine.If determining in decision steps 1515 that current 2D grid segmentation comprises comprises a little, then treatment scheme marches to step 1535, wherein interrupts the 2D trellis traversal comprising the frontier point place.Follow step 1535,2D grid (polyline) traversal can continue across other non-comprising or the unshielding summit.

If definite current segmentation does not comprise the OB point or comprises a little in decision steps 1515, then treatment scheme marches to step 1520 to determine whether SEG has the profile summit that single order is looked from looking the unit.In some embodiments, if the summit corresponding to current wedge and single order from the cross-shaped portion of the silhouette edge looking the unit and look then this summit is defined as the profile summit.This point is compound-contoured summit (CSV), and it is current wedge and is incident on from looking the cross-shaped portion of the wedge on the silhouette edge of looking the unit.

If determine in current segmentation, not exist the profile summit in decision steps 1520, then treatment scheme marches to step 1525, and wherein the 2D trellis traversal is filled traversal with the breadth first traversal of 2D grid or flood and marched to current segmentation and share the segmentation that the next one on non-profile summit does not travel through.

If current segmentation has the profile summit, then the normal breadth first traversal of 2D grid or flood are filled traversal not in the continuation of place, profile summit.Alternatively, if determine that in decision steps 1520 current segmentation comprises single order profile summit, then treatment scheme marches to step 1545 to come wedge line (WL) on the structural configuration summit (CSV) by rotation to the unity element (VCE) of looking that supports current wedge.

If the current wedge of processing is the SV-ME wedge, then the WL in the step 1545 is by coming WL the constitution step 154 from CSV rotation to the summit, unit of looking of constructing current wedge thereon.Discussing the purposes that is used for identification CSV and this processing of the wedge line that is configured to the SV-ME wedge shown in Figure 41 A and Figure 41 B and in conjunction with these figure in detail.In some embodiments, if the current wedge of processing is the SE-MV wedge, then by coming the WL the constitution step 1545 from CSV rotation to (inside corner) grid profile summit (or other CSV) of constructing current wedge thereon.

In other embodiments, in the situation that the grid silhouette edge with support that to look element sides parallel, if the current wedge of processing is the SE-ME wedge, then by constructing WL from CSV rotation to the support summit that belongs to the support of constructing current wedge thereon and look element sides.The purposes of identifying CSV and this processing of the wedge line that is configured to the SE-ME wedge at length is being discussed shown in Figure 41 C and in conjunction with this figure.

In follow-up decision steps, determine whether SE-MV type of the current wedge of processing.In this case, by rotating the WL that comes structure correspondence in step 1545 to inside corner grid profile summit.If determine that in decision steps 1550 the current wedge of processing is the SE-MV wedge, then treatment scheme marches to step 1553, otherwise treatment scheme directly marches to step 1555.

In step 1553, element sides is looked in the support that WL is extended to (SE-MV) wedge.In some embodiments, the cross-shaped portion of the support of WL and wedge being looked element sides is stored as the SVV for this wedge.This point is used for the final structure of initiation meeting scanning current wedge and the SE-MV wedge that is connected with the crossing wedge of current wedge at the CSV place of structure WL.If the current wedge of processing is the SV-ME wedge, then SVV is the SVV that is used for this wedge of structure of wedge.If the current wedge of processing is the SE-ME wedge, then SVV looks one of summit of element sides for this wedge of structure.In both cases, simplified determining that the clear and definite SVV that supports WL is ordered.

This SVV is as " beginning " SVV that scans processing of the SE-MV wedge that the current wedge of processing is connected with the wedge that is incident on CSV for final generation.These " assist " the connection wedge is for example to generate in the step 1240 of Figure 12 (enforcement of uncontinuity grid) and in the step 2060 of Figure 20 (directly constructing observability figure by the 3D trellis traversal).

In subsequent step 1555, the WL that generates in step 1545 and mesh triangles shape/segmentation (all intersect with current wedge) intersect to seek and crossing nearest from the some triangle of looking visible of current wedge.In one embodiment, adopted and accelerated the technology that this ray projection is determined, comprised that foregoing diabolo grid data carries out the hierarchical space segmentation.

Cross-shaped portion between the nearest mesh triangles shape of WL and identification in step 1545 be the conservative umbra uncontinuity grid of 2D with from supporting observability that looking of WL look on summit, unit (or the grid profile summit in the SE-MV situation) corresponding summit or point (2DDM_POINT).Because the profile summit is from looking to look in the unit rather than looking from looking unity element, so result's 2D uncontinuity grid is guarded.

The treatment scheme row marches to decision steps 1563 from step 1555 and is blocked to determine that whether " 2DDM_ point " look from corresponding VCE every other " part " except the summit of supporting WL.

If constructing the wedge of this 2D uncontinuity grid thereon is the SV-ME type, then because VCE self is point, so " 2DDM_ point " is VCE from correspondence look " otherwise being blocked ".Again, in the present context, " being blocked in addition " means that " 2DDM_ point " is to look from the every other part except the summit of supporting corresponding WL of VCE to be blocked.This also sets up for " 2DDM_ point " corresponding with the SE-MV wedge.

In both cases, because source " district " is a little, so " 2DDM_ point " is corresponding to blocking border (OB) point about the observability figure that looks from " district " of wedge.

In the situation that the SE-ME wedge, still, VCE is actual zone (line segment).In this case because the OB point be on the corresponding 2D observability figure expression " from the district look (VCE) be blocked " and " from the district look (VCE) be not blocked " between the point on border, block border (OB) point so " 2DDM_ point " needs not to be.

In Figure 16, provided in the processing details shown in the step 1563 and in conjunction with this figure it has been carried out detailed discussion.

If determine that in decision steps 1563 " 2DDM_ point " is to look from VCE otherwise to be blocked, and then adds this point to triangle/segmentation that (nearest) intersected as the OB point in step 1568.Otherwise " 2DDM_ point " do not added to this segmentation as the OB point and process from step 1563 march to step 1580 with determine current " traversal ", whether to exist any untreated/segmentation that do not travel through.In this case, (2D) " traversal " is defined as being connected to by non-profile summit a set of segmentation of current segmentation.

If determine to have the not segmentation of traversal in decision steps 1580 in traversal, then treatment scheme marches to step 1525, and the segmentation that wherein next one is not traveled through is processed.

On the other hand, if in decision steps 1580, determine in traversal, not exist untreated segmentation, then treatment scheme marches to step 1583, and wherein restriction comprises that segmentation that the OB that generated by current traversal order is effectively to remove the be blocked part of side and by this segmentation of OB point termination of correspondence of being positioned at of segmentation " looking from VCE " that OB orders.

Be added into these OB points of segmentation and be with for example identify during the 2D trellis traversal that comprises this fragment in step 1515 subsequently block borderline phase with block the border.

Treatment scheme marches to step 1585 with the 2D trellis traversal on the side that is not blocked that continues OB and order from step 1583.Because from the WL of a triangle/segmentation stream shape can be on different (unconnected) triangles/segmentation stream shape the OB point, the stream shape of the new connection that this " shelter fusions " that is illustrated in two stream shapes that the WL by OB point place forms forms upward (polyline polymerization umbra or PLAU) continuation travels through.

In a kind of enforcement of this method, the whole set of triangular mesh object is comprised by the mesh object of sealing.The mesh object of this sealing similar be generally used for around " day sylphon " for the mesh object of the visual application of interactivity.This grid traversal method among the 2D (and relevant 3D trellis traversal method) will seal day sylphon and be considered as any other triangular mesh stream shape.The method guarantees to travel through the omnidirectional's traversal that has realized whole database, even when do not present other mesh objects in database.Use this sealing grid so that current method has output susceptibility, this is because only need to process the grid that is not blocked that runs into, and does not need each grid in the process database necessarily.In addition, this method is identified the district that is not blocked of day sylphon itself naturally, since its size and position, its source of usually or significantly overdrawing.

Figure 15 and the relevant method that illustrates the observability figure that structure looks from the district, wherein the segmentation of each generation of the umbra uncontinuity grid of correspondence is tested with the segmentation of determining the observability figure whether this segmentation looks from the district (that is, from looking the border of blocking that the unit looks).The embodiment that substitutes processes to construct the uncontinuity grid segmentation that is produced by the wedge that is created on the silhouette edge that runs into 3D and 2D trellis traversal.In this embodiment that substitutes, whether the uncontinuity grid regions of then each result's uncontinuity grid regions being tested to determine this result travels through this limited shadow zone, or not limited non-this shadow zone of traversal.Should determine to adopt polyhedral some pardon test of the modification of using single order wedge or high-order UBP.

Figure 41 A shows with the method for Figure 15 and relevant figure to determine figure about the processing of the observability of wedge for simple SV-ME wedge.

Figure 41 B also shows for the SV-ME wedge and determines figure about the processing of the observability of wedge.In this case, wedge and polygonal mesh intersect so that result's profile summit is the tip of single order profile contour.

Figure 41 C shows with Figure 15 and relevant figure to determine figure about the processing of the observability of wedge for the SE-ME wedge.

In the discussion of relevant figure, provided details.

Figure 16: be used for determining whether 2D uncontinuity net point looks the processing that unity element (VCE) is looked and otherwise conservatively is blocked from the correspondence of wedge.

During main 2D trellis traversal shown in Figure 15 was processed, step 1563 was the judgement tests whether definite " 2DDM_ point " is looked and be blocked from the every other part of the VCE of wedge.Figure 16 shows a kind of embodiment of details of implementing this test.

The VCE of wedge is point or the line segment of determining about the observability of wedge.According to convention, " 2DDM_ point " looked from the strong point of the correspondence of VCE and is blocked.In fact, " 2DDM_ point " is on the border of looking from the corresponding point of VCE between being blocked and not being blocked.

If VCE self is a single point (SV-ME or SE-MV type wedge), then look is blocked corresponding " 2DDM_ point " from " all parts " of VCE, and this is that VCE does not have other parts because as a point.

In some embodiments, processing shown in Figure 16 starts from decision steps 1610, wherein determines whether SE-ME wedge of the wedge corresponding with " 2DDM_ point " that will test.

If determine that in decision steps 1610 corresponding wedges are not the SE-ME types, then treatment scheme marches to step 1615 and otherwise is blocked " 2DDM_ point " is arranged to look from VCE.As shown in Figure 15, step 1563 and step 1568, this is so that " the 2DDM_ point " tested becomes the OB point.

If determine that in decision steps 1610 corresponding wedges are SE-ME wedges, then treatment scheme march to step 1620 with at " the 2DDM_ point " tested be to look formation 2D axle (2DDM_VCE axle) between the VCE on limit of unit in this case.

Treatment scheme marches to (by the 2D traversal) triangle/segmentation that whether has any not traversal in the 2DDM_VCE axle that decision steps 1625 constructed to determine in step 1620.

If in decision steps 1625, determine in the 2DDM_POINT-VCE axle, to exist untreated triangle/segmentation, then process and march to step 1630, wherein the 2D traversal jumps to nearer untreated triangle/segmentation in " 2DDM_ point-VCE " axle, and processing shown in Figure 16 stops.

On the other hand, if in decision steps 1625, determine at 2DDM_ point-VCE " the not triangle segmentation of traversal in the axle; then treatment scheme marches to step 1635 with identification 2DDM_ point-VCE " all WL in the axle, and the WL of each identification is placed in the tabulation that is called LIST.In embodiment, in main 2D traversal step 1545 shown in Figure 15, these WL have been constructed.

Treatment scheme marches to decision steps 1640 with each wedge line of processing identification step 1635 and places it among the LIST to determine WL whether single order wedge line or high-order WL from 1635.With in conjunction with Figure 17 and Figure 18 the processing of adjusting single order WL with the high-order WL of formation back projection being discussed at length in this manual.

For each WL that in decision steps 1640, processes in the tabulation, if determine that handled WL is high-order WL, then treatment scheme marches to step 1645, the high-order WL that wherein in step 1640, identifies and be arranged in 2DDM_ point-axle " and the every other WL that is formed on the axle between WL and the VCE intersect.In some embodiments, the cross-shaped portion of this segmentation of high-order WL and other WL that intersect has potentially produced from the counterpart of the umbra boundary line (UBL) that VCE looks.Although WL reflected from the observability on the single summit that VCE looks, corresponding UBL reacted from whole VCE(in this case VCE look element sides) observability (be blocked or be not blocked) of looking.UBL can have the segmentation of looking and being blocked and looking and be not blocked from VCE from VCE, and above-mentioned segmentation is based on the WL of UBL and the cross-shaped portion of other WL.

Treatment scheme marches to decision steps 1647 to determine whether any untreated WL is arranged " tabulation " from step 1645.If untreated WL is arranged in " tabulation ", then process and return step 1640.

If determine that in decision steps 1640 handled WL is single order WL, then treatment scheme directly marches to step 1647 to determine whether remain any untreated WL in " tabulation ".If untreated WL is arranged in " tabulation ", then treatment scheme is returned step 1640.

If determining " tabulation " in decision steps 1647 be sky, then treatment scheme marches to step 1650, and whether the polygonal some pardon that wherein makes " 2DDM_ point " to stand to revise is tested to look from VCE with definite its and be blocked.As discussing in embodiment in front, in conjunction with the step 1445 among Figure 14, the 2D version of polyhedral some pardon test of modification shown in Figure 25 is adopted in this test, in conjunction with this figure it is discussed.Single order event face (single order wedge) is used in polyhedral some pardon test of this modification, or the wedge line of the correspondence in the 2D situation (WL), and alternatively, is high-order umbra boundary polygon (UBP), or the umbra boundary line (UBL) of the correspondence in the 2D situation.This test allows needn't construct whole polyhedron/polygon to the implicit solution of polyhedral some pardon test or polygonal some pardon test.

Treatment scheme marches to step 1655 with the result of polygonal some pardon test of this modification of inspection step 1650 from step 1650.If " 2DDM_ point " is blocked, then treatment scheme marches to step 1660 and to return is, otherwise that treatment scheme marches to is 1665 no to return.

Figure 17: 2D trellis traversal: construct the high-order wedge by back projection

At Figure 14, Figure 15 and Figure 16 and the relevant described 2D trellis traversal processing employing single order wedge line (constructing in the step 1454 of Figure 15) for determining about the observability of wedge of figure.

In the situation that SV-ME wedge and SE-MV wedge, corresponding wedge line is the visibility events face of looking from point accurately.In these cases, because source " district " is degenerated to a little, so WL or for the umbra boundary line (UBL) of looking from the district of the visibility problem of " looking from the district ".

But, in the situation that the SE-ME wedge, about the actual visibility problem of looking from the district of determining of the observability of wedge, wherein determine observability according to looking the unit.In this case, in some embodiments, (be actually degree of freedom and be 1 face: wedge line) only guarantee conservatively approximate corresponding accurate event face, they need not to be identical line to be configured in single order visibility events face on the summit of looking the unit.

This is to look as seen from the profile summit of structure wedge line because the structure of single order WL is supposed whole source or vision area (in this case for looking element sides).In this case, because this hypothesis always sets up for single viewpoint, so the first order modeling that the observability of developing in polyhedron/polygon environment in the present embodiment is propagated is considered as vision area " viewpoint " in some respects.

But for vision area, this hypothesis may be false.In fact, when observing from the profile summit of supporting WL, looking summit, unit (SVV) for the support of structure WL in fact may be blocked.In this case, accurate WL is not identical with single order WL.

In some embodiments, in order to determine to be incident on more accurately the WL on the identical single order profile summit, the back projection processing of looking from point is adopted on single order profile summit as viewpoint.In back-projection approaches, certain structure of using grid is as looking source, unit (being single order profile summit) in the 2D situation, in vision area from it (in the observability situation of this 2D about wedge, being to look element sides) structure uncontinuity grid or observability figure.

Summit, unit (VSVV) is looked in conservative (looking from the profile summit) visible support that this method has adopted focusing conservative back-projection approaches to construct WL thereon with identification.Should " adjustings " only be applied in following situation to WL: look summit, unit (SVV) for the support of the structure of original orthogonal projection single order WL and look from the support profiles summit and be blocked.Otherwise single order event face (WL) is accurate.

Embodiment adopts the 2D trellis traversal method of calculating back projection, with similar in the method shown in Figure 14, Figure 15 and Figure 16 about the situation of the observability of wedge for the orthogonal projection (looking from looking unity element) about wedge.

The back projection problem about wedge of seeking VSVV often is the visibility problem of looking from point, and this is because often be a little about looking from VCE contour structure (rear-projection eikonogen) of wedge.

In some embodiments, identification about wedge, from the profile summit look VSVV processing in the situation that SVV be blocked from selecting step 1705(Figure 17 of seed mesh triangles shape/segmentation).

2D trellis traversal for back projection: the method that only adopts the triangle/segmentation of orthogonal projection-be not blocked

In some embodiments, in back projection, only use visible those triangle/segmentations orthogonal projection (looking from looking element sides).In the method, visible triangle/segmentation makes their planar orientation be calculated as oppositely with respect to back projection in orthogonal projection.Usually, when observing for the grid profile summit of " source " point of back projection from conduct, orthogonal projection profile summit is identical summit with back projection profile summit.In this case, before the application-oriented counter-rotating, there is not the segmentation of orthogonal projection towards back projection source point (single order profile summit)-be not blocked.

In other cases, the visible grid elements of not every orthogonal projection can be with respect to the re-projection source point back side pair.In these cases, if the orientation of triangle/segmentation is inverted, then its can be with respect to the back projection source point back side pair.This is so that it with respect to this back side pair, therefore, has created the grid that triangle/segmentation has inconsistent orientation.In this case, will develop with respect to back projection point is that the back side removes from grid the triangle/segmentation that is orientated.

Fail to be included in the precision of the VSVV that the geometric figure that blocks potentially in the back projection may reduce to calculate and the relevant adjustment of WL.But, still guarantee the result that guards, this is because any geometric elimination of blocking potentially only can be so that the VSVV that calculates be the more conservative estimation to SVV.In addition, by only adopting orthogonal projection-visible triangle/segmentation, reduced significantly the cost that back projection is processed, this is because relevant orthogonal projection-visible triangle/segmentation has determined when processing for specific grid profile summit initiation back projection.

In the method, in fact the visible grid elements of orthogonal projection is used as calculating the virtual shelter of back projection.

Therefore in step 1705, make the selection of seed 2D grid according to visible (not being blocked) grid collection in orthogonal projection.

2D trellis traversal for back projection: the retraction on high-order observability bottom profiled summit

In a kind of embodiment of this method, back projection is only definite by those rear-projection eikonogens, and in this case, above-mentioned rear-projection eikonogen is grid profile summit, and when observing from VCE, it is the single order profile summit in the orthogonal projection.In this is implemented, only those WLS that are attached to single order, orthogonal projection profile summit are carried out " adjusting " with the observability of reflection high-order about the plane by back projection.The WL that regulates keeps being attached to original grid profile summit.

In fact, when any summit as the single order of observing from SVV, orthogonal projection profile summit is when observing from VSVV, because the triangle/segmentation on shared this summit can be with respect to the VSVV back side pair, so this summit no longer is the profile summit.In this case, orthogonal projection profile summit (rear-projection eikonogen) in fact look from looking unity element (in this case for limit) be blocked.In this case, the umbra event face that actual (accurately) looks from the district is not supported by or not this orthogonal projection profile summit, and identical 1-stream shape is upper supports apart from looking nearer summit, unit but be positioned at.Single order is the potential effect of the high-order observability determined by back projection from the profile looking the unit and look towards the migration of looking the unit or " retraction ".In some cases, fail to explain the precision that the correspondence of VSVV that this profile migration may reduce to calculate and WL is regulated.But, because compare the corresponding event face of the profile that is attached to corresponding retraction, the visibility events face (being WL in this case) that is attached to the adjusting on orthogonal projection single order profile summit always can produce more conservative umbra event face, and all have still guaranteed conservative result.

Be attached to the WL of the adjusting on its original single order orthogonal projection profile summit by maintenance, " the support segmentation " of the WL of adjusting between profile summit and VSVV can be violated " the local observability " that it stays the grid of depositing potentially by penetrating grid surface.Accurately, the profile summit can be easy to towards looking the unit migration so that produce adjusting WL and such summit of not violating local observability of the nearer connect Vertex of grid become " accurately " profile summit for a moment in this case.

WL for the conservative adjusting that keeps being attached to its original single order orthogonal projection profile summit becomes the back projection Treatment Design violation that prevents local observability by the WL that regulates.This occurs in, and the WL that works as adjusting originates from orthogonal projection single order profile summit so that WL is positioned at the inside of mesh triangles shape 1-stream shape (polyline).This is that 1-flows a class of shape from blocking, it can be by to detecting to prevent with all grid elements (be the ambiguity line segment in this situation) that orthogonal projection profile summit forms the assembly is connected, wherein the complete inside of the axle between the segmentation of VCE limit and summit of the assembly of connection.This can carry out with the simple traversal to polyline that begins from this summit and finish when intersect on segmentation and axle border.These grid elements are the grid elements from type of barrier that can produce the retraction that causes the profile summit.By determining to identify according to back projection and eliminating these elements, back projection can produce conservative " adjusting " WL, and but its maintenance is attached to original orthogonal projection profile summit has explained so that the high-order observability effect that SVV looks and is blocked from this profile summit.

Therefore, in the step 1705 of Figure 17, do not consider namely can cause potentially for the seed 2D grid elements of back projection the selection of those grid elements of certainly blocking on back projection-summit, source.

The method simplified the 2D trellis traversal about the enforcement in this situation of the observability of wedge.When the 3D trellis traversal adopt the method about the observability situation (as discussing in conjunction with Figure 19, Figure 20 and Figure 21) of looking the unit is lower the time in addition can realize larger simplification.By adopt the method in 3D trellis traversal situation, event face (wedge polygon) keeps being attached to original single order orthogonal projection profile contour to guarantee to construct continuous, the conservative umbra boundary surface of looking from the district.

Again, following alternative embodiment can be arranged: the retraction that the explanation of certainly blocking is made it possible to carry out the profile summit.For example, the method on a kind of searching " retraction " profile summit is to calculate back projection and determine whether the WL corresponding with the VSVV that calculates violates local observability according to initial orthogonal projection, single order profile summit.If violate local observability, then when considering the high-order effect this summit in fact from look the unit look invisible and will be apart from the VCE limit nearer adjacent vertex as for the source point of back projection to determine whether the WL corresponding with the VSVV that newly puts violate local observability.Can repeat this processing until run into the WL that does not violate local observability.With the high-order WL of this WL as adjusting.

Usually, although the back projection complexity that increases can be clearly, only produce the additional amount of blocking of looking from the district by the retraction on profile summit usually very little.Remain fixed to the orthogonal projection WL on original orthogonal projection, single order profile summit and calculate " adjusting " WL corresponding with VSVV by back projection, back projection is to have explained a kind of method of explaining most of high-order observability effects with very little increase cost.

Emphasize such as the front, only just use the adjusting of high-order back projection when being blocked when original SVV looks from single order profile summit.In some embodiments, adopt the simple maximum possible deviation of measuring between the high-order WL that determines single order WL and corresponding adjusting.For example, by single order WL(namely, Support Level) and profile summit and angle that corresponding defiber between the VCE forms be maximum angle deviation between single order WL and the high-order WL; Because use the method, if VSVV is burble point, then high-order WL can only be equal to defiber under limiting case.This angle is easy to increase along with the increase of the distance of distance VCE naturally.That is, vision area seems more " point-like " from rear-projection eikonogen.Thereby, determine that by back-projection approaches high-order observability effect becomes not too important in the larger distance of distance vision area.In some embodiments, measurement of angle is used for determining whether and calculating the high-order effect for any given projection, single order profile summit.

Now return Figure 17, it shows for the overall control of initiating, again initiating and stopping 2D back projection traversal and processes, and selects the seed triangle/segmentation of some initial 2D grids about wedge to initiate traversal shown in 1705.

As in orthogonal projection is implemented (step 1405 of Figure 14), this processing has been accelerated in the selection that is likely the initial seed triangle/segmentation of exposing shelter.

Treatment scheme marches to step 1710 to throw comprising a little on the structural species sub-grid with ray from 1705.About the back projection problem of wedge be limited in by back projection source point (grid profile summit) and VCE(in this case look element sides) district in the 2D axle that forms.Therefore, in some embodiments, comprising a little of identification step 1710 come on the limit that is used for the ray of ray projection with the conduct of this axle.

Treatment scheme marches to step 1715 and processes to call for the main of the 2D trellis traversal method of looking from the summit of the visible segmentation of determining this axle.

As main process (shown in Figure 15) of the 2D trellis traversal method of looking from the limit, the main processing of the back projection trellis traversal of looking from point can cause termination (for example termination of the step 1830 of Figure 18).Treatment scheme marches to decision steps 1720 to determine whether to exist to come the main termination traversal of processing (Figure 18) of the back projection 2D trellis traversal of looking since the summit.

Because the back projection Treatment Design is at original source VCE structure observability figure, so step shown in Figure 17 and step shown in Figure 14 similar (orthogonal projection, construct about the observability figure of wedge).Similarly, except source in the back projection situation be the summit and in the orthogonal projection situation source be limit (SE-ME wedge situation), the described step of Figure 17 (back projection, about the observability figure of wedge) and step shown in Figure 180 (orthogonal projection, about the observability figure of wedge) are similar.

Process flow diagram shown in Figure 17 shows initiation, again initiates and stop at the process flow diagram of the processing of the main 2D traversal processing that is shown specifically among Figure 18 and describes in detail at the further part of this instructions.

If the segmentation of determining not exist the major part of 2D trellis traversal to end in decision steps 1720, then treatment scheme marches to step 1725, and wherein the 2D trellis traversal for wedge is terminated, thereby has finished the structure of 2D about the observability figure of wedge.

If determine the segmentation (in the step 1830 of Figure 18) that exists main 2D trellis traversal to end in decision steps 1720, then treatment scheme marches to decision steps 1730 to determine by ending whether there is any triangle in the axle that segmentation (SEG) and rear-projection eikonogen summit form.

If determining in decision steps 1730 does not have triangle in by the axle between SEG and the summit, source, then treatment scheme marches to step 1755 to construct comprising a little on the 2D grid (polyline) that is formed by the segmentation that is connected to current termination segmentation.

Treatment scheme marches to step 1760 again to initiate the 2D trellis traversal to previous termination segmentation.

If determine to have triangle in decision steps 1730 in by the axle between SEG and the summit, source, then treatment scheme marches to step 1735 to determine whether any not triangle/segmentation of traversal is arranged in the SEG-VCE axle.

If in decision steps 1735, determine in " SEG-summit " axle, to exist the not triangle/segmentation of traversal, then treatment scheme marches to step 1740, wherein " redirect " or restart the 2D trellis traversal in the triangle/segmentation of the nearer not traversal in " SEG-summit " axle.Appear at also that this step in the back projection master 2D traversal processing step 1830 shown in Figure 180 was guaranteed before this traversal continues and stop having handled triangle/segmentation that all block potentially before the determining of traversal (because end segmentation look from the summit, source blocked fully) making.

If determine not exist not the triangle of traversal in " SEG-summit " axle in decision steps 1735, then treatment scheme marches to step 1745 to determine whether point in current (terminations) segmentation is positioned at the look any polyline polymerization umbra (PLAU) in one or more polygons subarea of being blocked of the summit, source from wedge that represents wedge.In some embodiments, the 2D version of polyhedral some pardon test of modification shown in Figure 25 is adopted in this test.Single order event face (single order wedge) is used in polyhedral some pardon test of this modification, or the wedge line of the correspondence in the 2D situation (WL), and alternatively, high-order umbra boundary polygon (UBP), or the umbra boundary line (UBL) of the correspondence in the 2D situation.This test allows needn't construct whole polyhedron/polygon to the implicit solution of polyhedral some pardon test or polygonal some pardon test.

If determine that in decision steps 1745 ending segmentation point is positioned at PLAU, then treatment scheme marches to step 1750, wherein travels through not restart and end segmentation in the termination segmentation no longer to be marked as termination.Treatment scheme stops at 1750 places.

Be not positioned at any PLAU if determine to end segmentations in decision steps 1745, then treatment scheme marches to step 1755, and wherein the 2D grid that is associated of structure comprises a little.Treatment scheme marches to 1760 again to initiate as previously mentioned to travel through in conjunction with the step 1460 of corresponding Figure 14.

Figure 18: back projection, the 2D trellis traversal master who looks from the summit process

According to some embodiments, process similar with shown in Figure 15 about wedge, the back projection 2D trellis traversal of looking from the summit about main processing that wedge, the back projection 2D trellis traversal of looking from the summit are processed.

In some embodiments, processing shown in Figure 180 starts from step 1805, and it is to begin 2D grid (polyline) traversal about the mesh triangles shape segmentation of wedge to the traversal that does not travel through segmentation.Polyline traversal from segmentation to segmentation is forward, advances along one of both direction from segmentation to adjacent sectional.The corresponding 3D traversal (describing in detail in the discussion of Figure 20) of triangular mesh can be passed one of three adjacent boundaries and be occured, and for example advances with flood filling or breadth first traversal.

Treatment scheme marches to step 1810 to determine whether to exist in the 2D axle between current segmentation and summit, source the triangle/segmentation of any not traversal (processing by current 2D trellis traversal).Single initial wedge and the cross-shaped portion of a plurality of visible mesh triangles shapes can produce a plurality of ambiguity line segments and a plurality of corresponding 2D axles potentially.In a kind of enforcement, construct 2D axle (" SEG-summit " axle) in these segmentations each.Following alternate embodiments is possible: the axle that the single conservative axle that surrounds the segmentation of a plurality of connections is used for accelerating step 1810 comprises test.If have any not triangle/segmentation of traversal in the axle in larger comprising, then process each axle that to get back to for each segmentation.

In some embodiments, by adopting hierarchical space segmentation (and/or bounding volume hierarchy (BVH) structure) tissue 3D polygonal mesh geometric figure to come the axle of further accelerating step 1810 to comprise test.The cost that any hierarchical organization in usage data storehouse is easy to axle is comprised test is reduced to about NLog (M) from about N*M, and wherein M is the polygonal quantity in the model, and N is the quantity of the axle tested.

If there is the not segmentation of traversal in " SEG-summit " axle, then treatment scheme proceeds to step 1830, and wherein the 2D traversal jumps to the nearlyer segmentation in " SEG-summit " axle.These two steps are carried out vertical processing order.

If determine not exist the not segmentation of traversal in " SEG-summit " axle in decision steps 1810, then treatment scheme marches to step 1815 to determine whether the current segmentation that is traveled through comprises back projection and block border (BOB) point or the no frontier point that comprises.

In some embodiments, back projection border (BOB) some expression 2D ambiguity linear flow shape is upper looks visible and the border between being blocked of looking from the summit from the summit.The BOB point appears at the polygonal mesh segmentation cross-shaped portion place on back projection wedge line (BWL) and the wedge.BWL and each cross-shaped portion of polygonal mesh segmentation are that corresponding 2D(is about wedge) the 1D element (point) of uncontinuity grid.

If determine that in decision steps 1815 current segmentation comprises the BOB point, then treatment scheme marches to step 1835 to stop current 2D trellis traversal at BOB point place, and wherein processing shown in Figure 180 is interrupted.In other embodiment, treatment scheme marches to from step 1835 and processes 18-1, and it is back to step 1825 with treatment scheme.

Similarly, determine in decision steps 1815 whether the segmentation of current 2D grid comprises the containing border.This frontier point be 2D grid and wedge such as the processing in the main back projection 2D traversal processing of the step 1710 of Figure 17 or 1755(control Figure 17) in the cross-shaped portion that defines boundaries.If determining in decision steps 1815 that the segmentation of current 2D grid comprises comprises a little, then treatment scheme marches to step 1835, wherein interrupts the 2D trellis traversal containing the frontier point place.In other embodiment, treatment scheme marches to from step 1835 and processes 18-1, and it is back to step 1825 with treatment scheme.If definite current segmentation does not comprise the BOB point or comprises a little in decision steps 1815, then treatment scheme marches to step 1820 to determine whether current segmentation comprises the profile summit.

Because rear-projection eikonogen is summit (point), the determining of step 1820 is equal to the profile of looking from point of seeking the place, summit that is positioned at current segmentation.

If in decision steps 1820, determine in current segmentation, not exist the profile summit, then treatment scheme marches to step 1825, and wherein the 2D trellis traversal is used the breadth-first of 2D grid or flood to fill traversal to march to current segmentation and share the segmentation that the next one on non-profile summit does not travel through.

If current segmentation has the single order point, then the normal breadth-first of 2D grid or flood are filled traversal in the continuation of place, profile summit.Alternatively, if determine that in decision steps 1820 current segmentation comprises single order profile summit, then treatment scheme marches to step 1845 and constructs the back projection wedge line (BWL) that is incident on the single order profile summit to look the unit by rotation to corresponding support.

In step 1845, wedge line (WL) is by being rotated to form to the line with back side segmentation (as defined above) conllinear, and wherein rotation is along carrying out away from the direction on the polygonal outside of correspondence to summit, source.

Treatment scheme marches to step 1855 from step 1845, and the WL that wherein generates in the step 1845 intersects with the mesh triangles shape/segmentation of intersecting with current wedge, with seek that nearest and current wedge intersect from the some triangle of looking visible.In one embodiment, can utilize BWL to determine this cross-shaped portion as ray with the ray projection.In some embodiments, adopt the aforesaid definite technology of this ray projection of acceleration that comprises the hierarchical space segmentation of triangular mesh database.

BWL with in step 1855 cross-shaped portion between the nearest mesh triangles shape of identification be summit or point (" 2DDM_ point ") with the conservative umbra uncontinuity grid of 2D corresponding to the observability of looking from the summit, source.Because this is the visibility problem of looking from point, so such " 2DDM_ point " all is or the BOB point.

Treatment scheme marches to decision steps 1880 to determine whether there is any being untreated/not segmentation of traversal in current " traversal " China from step 1855.In this case, (2D) " traversal " is defined as a set of segmentation that is connected to current segmentation by non-profile summit.

If determine to have the not segmentation of traversal in decision steps 1880 in this traversal, then treatment scheme marches to step 1825, and the segmentation that wherein next one is not traveled through is processed.

On the other hand, in this traversal, there is not untreated segmentation if in decision steps 1880, determine, then treatment scheme marches to step 1883, and wherein restriction comprises that segmentation that the BOB that generated by current traversal order is effectively to remove being positioned at of segmentation " looking from the summit " that BOB orders part on the side and this segmentation BOB point termination by correspondence that is blocked.

These BOB points that are added into segmentation be with following block borderline phase with block the border: that for example identifies during the traversal to the 2D grid that comprises this segmentation in step 1815 subsequently blocks the border.

In subsequent step 1885, the side that is not blocked of ordering at BOB continues the 2D trellis traversal.Because the BWL from a triangle/SEG stream shape may be different (unconnected) triangles/SEG stream shape produces the OB point, so this expression is to the continuation of the traversal of " shelter fusion " the formed new connection stream shape (back projection polyline polymerization umbra or PLAU) of the BWL that passes through BOB point place of two stream shapes.(for the description that the shelter among the 3D merges, referring to the people such as Durand (2000), its full content merges among the application by reference)

Figure 19: 3D trellis traversal

Refer now to Figure 19 and relevant Figure 20 and Figure 21, process (traversal of 2-stream shape) from the 3D trellis traversal of the observability figure that looks the unit and look and be described being used for the structure single order.

The visibility problem that the 3D trellis traversal has dealt with 3D, looked from looking the unit.The output that the 3D trellis traversal is processed by use conservative linearization umbra event face (CLUES) structure from looking the unit linearization umbra observability figure that guards that the polygon segmentation that is blocked forms that looks.As described, in some embodiments, CLUES is corresponding to conservative visibility events face or " wedge " of linearization according to the method construct of the exemplary process diagram of Fig. 1 and Fig. 3 to Fig. 6.

Be different from used wedge in the uncontinuity gridding method of prior art, the CLUES that adopts in the method or single order wedge are based on that conservative model that the single order observability propagates limits.In single order observability method, the contour feature by polygonal mesh and look the structure that element characteristic limits wedge only; Being inserted in geometric figure between these two structures does not affect the structure of single order wedge.

The overall tissue of 3D trellis traversal method is (for example to flow the shape grid elements, the triangle of stream shape triangular mesh) traversal is at first identified all grid elements that can completely or partially block the grid elements that is traveling through to guarantee for any grid elements that is traveling through.Before the traversal of the original mesh element that traveling through is carried out, at first process (recursively processing by the 3D trellis traversal) these grid elements of blocking potentially.During the 3D trellis traversal is processed, construct the umbra event face that comes from these potential grid elements of blocking at the single order silhouette edge that does not block of the element that blocks potentially.These event faces can intersect with mesh triangles shape, cause may with can limit traversal processing block umbra uncontinuity corresponding to border from what look the unit looks.

This vertical processing order is limited in those wedges that are incident on the visible silhouette edge with the structure of visibility events face (wedge) to a great extent, and this has reduced to carry out the time quantum of trellis traversal.The method also is restricted to traversal the mesh triangles shape that is blocked to a great extent, is that output is responsive so that process actually from looking the look quantity of visible element of unit.According to some embodiments, partly block the element that is not blocked.

This is opposite with the conservative umbra uncontinuity irregular triangular mesh design method of single order shown in Figure 12, in method shown in Figure 12, and at all single order silhouette edge structure single order wedges, or even the single order silhouette edge that is blocked.In the insensitive method of the output of Figure 12, the uncontinuity grid regions that is formed by these single order silhouette edges that are blocked finally is confirmed as being blocked.But, because the event face is formed on all single order silhouette edges, so form all possible umbra uncontinuity grid regions and determine actually that subsequently the observability of looking from the district for reality checks above-mentioned uncontinuity grid regions in the post-processing step of layout of discontinuous grid.

In a kind of embodiment of this method, it is protruding supposing to look the unit.Suppose that also polygonal mesh is that triangular mesh and triangular mesh are stream shape (each limit are accurately shared by two triangles).The cross-shaped portion of mesh triangles shape and wedge or UBP can produce more complicated grid polygon.Can use the polygon triangulation polygon that these are more complicated of establishment to be transformed into triangle.(referring to M.de Berg, and M.van Dreveld et.al in " Computational Geometry Algorithms and Applications, Springer c.1997, page45, its full content merges among the application by reference).Triangular mesh originally is expressed as digraph.The method also is fit to polygon except being suitable for triangle, but uses triangle can simplify this traversal method.These hypothesis have been simplified enforcement, but other enforcements of relaxing these hypothesis also are possible.

The output that the 3D trellis traversal is processed is by from looking the unit conservative observability figure that grid polygon (with the polygon segmentation) forms that looks visible.These polygon segmentations are together with being incident on the border that comprises polyhedron polymerization umbra (PAU) from looking umbra boundary polygon (UBP) on the silhouette edge of looking the unit, it is the spatial volume of looking and being blocked from looking the unit, and the volume that wherein is blocked may conservatively be underestimated.Each 2-stream shape " fusion " extremely can be flowed in shapes from the UBP that more closely flows shape and farther other the crossing 2-of shape that flow, thereby realize blocking fusion.

In this 3D trellis traversal method, can construct clearly in the situation of the UBP corresponding with wedge shelter to the type and merge and identify.In the method, only must determine whether to look and be blocked from looking the unit corresponding to the uncontinuity grid segmentation of the cross-shaped portion of wedge and triangular mesh.If so, then this uncontinuity grid segmentation is the PAU segmentation, from but from looking the border of blocking that the unit looks.Compare and construct clearly all UBP(it needs wedge-wedge cross-shaped portion), the method has lower assessing the cost.

As shown in follow-up Figure 22, Figure 23 and Figure 24, UBP that will be corresponding with high-order or " adjusting " wedge clearly is configured to explain high-order observability effect.

The main initiation of processing of processing controls shown in Figure 19 3D traversal shown in 20, again initiate and stop.These processing are similar with control and main processing that Figure 14 and corresponding 2D shown in Figure 15 and that discuss in front travel through.

According to some embodiments, processing shown in Figure 19 starts from 1905 to select the seed triangle in kind of sub-grid.Select the grid of " strong as seen " and/or mesh triangles shape (at grid/triangle and look in the axle that the unit forms do not have other triangles) to accelerate overall traversal processing, still also can select arbitrary mess/triangle.

When having selected the seed triangle, treatment scheme marches to 1915 to initiate the main processing of 3D trellis traversal.In Figure 20, show in detail the main embodiment of processing of 3D trellis traversal.Generally, main 3D trellis traversal processing shown in Figure 20 is what to process to initiate by control shown in Figure 19.In addition, in some embodiments, main process (shown in Figure 20) of 3D traversal processing can be ended (step 2030) at grid specific triangles place.Processing shown in Figure 19 determines whether remain the traversal (step 1920) of any termination in main 3D stream shape traversal processing and main the processing in the segmentation that whether should end in these of 9D trellis traversal initiated (step 1930 and step subsequently) again.

If determine not exist the triangle of the major part of having ended the 3D trellis traversal in decision steps 1920, then treatment scheme marches to step 1925, wherein stops the 3D trellis traversal for wedge, thereby finishes 3D from the structure of the observability figure that looks the unit and look.

If determine to exist the triangle of ending main 3D trellis traversal (step 2030 of Figure 20) of 3D grid in decision steps 1920, then treatment scheme marches to decision steps 1930 to determine by ending triangle (TRI) and looking in the axle that the unit forms whether have any triangle.

If determine ending triangle and looking in the axle between the unit do not have triangle in decision steps 1930, then treatment scheme marches to step 1960 and again initiates the 2D trellis traversal with the segmentation to previous termination.Treatment scheme stops at 1960 places.

If determining in decision steps 1930 has triangle in ending triangle and the axle of looking between the unit, then treatment scheme marches to step 1935 to determine ending triangle-look any whether the triangle that does not travel through is arranged in the unit shaft.

There is the not triangle of traversal if in decision steps 1935, determine in the unit shaft ending triangle-look, then treatment scheme marches to step 1440, wherein in triangle/segmentation of ending triangle-the look nearer not traversal in the unit shaft " redirect " or restart the 3D trellis traversal.This step (its also in the step 2030 of main 3D traversal processing shown in Figure 20 occur) guarantee before traversal continues and make stop traversal (because the triangle of ending from look the unit look be blocked fully) decision handled the triangle that all block potentially before.

If in decision steps 1935, determine in the SEG-VCE axle, not exist the not triangle of traversal, then treatment scheme marches to step 1945 determining whether the point on current (terminations) triangle is positioned at any polyhedron polymerization umbra (PAU), and so looks and conservatively be blocked from looking the unit.In some embodiments, polyhedral some pardon test of modification shown in Figure 25 adopted in this test.

Single order event face (single order wedge) is used in polyhedral some pardon test of this modification, and alternatively, high-order umbra boundary polygon (UBP).This test allows needn't construct whole polyhedron for the implicit solution of polyhedral some pardon problem.Provide the details of this test in conjunction with Figure 25.

If determine that in decision steps 1945 point of ending on the triangle is positioned at PAU, then treatment scheme marches to step 1950, wherein again initiates traversal and no longer should end triangular marker for ending ending triangle.Treatment scheme stops at 1950 places.

If determine that in decision steps 1945 ending triangle is not positioned at any PAU, treatment scheme marches to step 1960, and it initiates traversal again to ending triangle.Treatment scheme stops at 1960 places.

If not having mesh triangles shape is the part of the coupling assembling of observability figure (PAU), then whole grid is blocked or is not blocked by VM/PAU by VM/PAU.Thereby, in some embodiments, in order to determine all polygonal observabilities of this wedge, with its Figure 19 and for example the method for Figure 25 determine the observability with respect to the coupling assembling of the concrete VM/PAU that finishes or correspondence of a grid vertex only.But, described in detail such as other places in this manual, by using 3D trellis traversal method in conjunction with the large bounding box that comprises all model geometric figures or sealing mesh object, wherein the bounding box polygon may be subdivided into less polygon, has naturally avoided the many tests in these tests.

Figure 20 A shows with the 3D trellis traversal and constructs Conservative Linear from the main process flow diagram of processing of the observability figure that looks the unit and look.

Line turns to Figure 20 A, and main 3D trellis traversal processes to scheme to fill the traversal beginning at breadth-first or the flood of the stream shape triangle shown in the initial step 2005 or other polygonal meshs.In processing, the control shown in the process flow diagram of Figure 19 selects the initial delta of the whole processing of beginning.

In one embodiment, grid is the stream shape triangular mesh that each limit is accurately shared by two triangles.Other definition of grid also are possible, comprise the grid that does not seal.The cross-shaped portion of mesh triangles shape and wedge or UBP can produce more complicated grid polygon.Can use the polygon triangulation method polygon that these are more complicated of establishment to be transformed into triangle.Triangular mesh is expressed as digraph.The method also is fit to the grid polygon except being fit to triangle, but uses triangle to simplify this traversal method.The stream shape triangular mesh that method shown in Figure 20 A adopts each limit accurately to be shared by two triangles.

In some embodiments, the treatment scheme shown in Figure 20 A has wherein been pointed out the traversal of mesh triangles shape since 2005.Traversal can breadth-first or flood blank map traversal order advance.Use the breadth first traversal of triangular mesh, traversal forms the traversing triangle layer from initial delta to external expansion.The triangle that is connected by non-silhouette edge can be said to " traversal " that belongs to identical.

Other traversal order are possible.Select not affect for the triangle of initiating one or more polygonal meshs the output of traversal.But, by selecting fully visible and near the initial delta of looking the unit, processing being easy to carry out better.In the processing that keeps the digraph traversal, step 2005 indication traversal is limited to the triangle that does not travel through in the grid.

Treatment scheme marches to decision steps 2010 to determine by current triangle with look any whether the triangle that does not travel through is arranged in the 3D axle (this axle is called as the TRI-VC axle) that the unit forms.If one or more not triangles of traversal are arranged in this axle, then treatment scheme proceeds to step 2030, and wherein traversal jumps to the nearer not triangle of traversal immediately.In one embodiment, organize the grid polygon with the hierarchical space sub-structure.This is so that the axle test of step 2010 can be by producing the previous traversal of the polygonal large grouping of grid promptly identify to(for) the LogN performance of axle test.

Step 2030 is carried out effectively for single mesh object or for the strict from front to back processing of the grid elements of a plurality of mesh objects.Traversal is jumped to the nearer leg-of-mutton processing of blocking potentially can be produced by the overlapping endless circulation that causes of leg-of-mutton circulation.In Weiler-Atheron observability algorithm, also run into such circulation, and can represent to detect and eliminate such circulation by the digraph of safeguarding overlapping relationship.In Figure 13, propose the method and in conjunction with this figure the method has been carried out detailed discussion.

The event face that step 2030 is guaranteed to come from the triangle of current leg-of-mutton triangle-the look nearer not traversal in the unit shaft was identified in the past and constructed in current leg-of-mutton traversal.These event faces can be to block border (it is determined for these nearer silhouette edges to step 2068 in step 2045) from what look the unit looks, and it has limited current leg-of-mutton traversal (step 2035).By carrying out the nearer not redirect of the triangle/silhouette edge of traversal, step 2010 is guaranteed to limit any of current leg-of-mutton traversal and was configured before current leg-of-mutton traversal from looking the border of blocking of looking the unit.Treatment scheme is ended at 2030 places, triangle-look in the unit shaft nearer, the mesh triangles shape place of traversal initiates again.

If the triangle of traversal not in the TRI-VC axle, then whether treatment scheme marches to decision steps 2015 and comprises from looking the border of blocking that unit (" from-VC ") looks with the current triangle of determining to travel through.Umbra uncontinuity grid (DM) segmentation generally forms polyline in the intersection of mesh triangles shape and wedge.Each DM segmentation is wedge and leg-of-mutton cross-shaped portion, and wherein segmentation is looked unity element (summit or limit) from the support of wedge and looked conservatively as seen.This is called as the visible cross-shaped portion of " about wedge ".Because be to use (the 2D of Figure 15 trellis traversal is processed) determined from looking the profile summit (rather than " looking unity element from the support of wedge looks " profile summit) of looking the unit about the visible segmentation of wedge, so about the visible segmentation of wedge conservatively as seen.Because segmentation in fact with respect to the support of wedge look unity element may the back side to rather than with respect to looking the unit just in the face of (being to look as seen from looking the unit therefore), so with respect to the support of wedge look the summit, unit in fact some right segmentations of the back side can be used as the DM segmentation and be included.

In step 2055 subsequently, come visible cross-shaped portion or the segmentation of uncontinuity grid about wedge are identified by Figure 15 and the relevant described 2D trellis traversal of figure.

DM_SEG comprises the polyline of conservative linearization umbra uncontinuity grid.These polylines can or can not be actual to block border (polygon segmentation Remote edge circle that will look conservatively visible polygon segmentation and look and be blocked from looking the unit from looking the unit, wherein the size of umbra volume is conservatively underestimated) from what look unit (" from-VC ") looks.Each DM_SEG blocks border (OB_SEG) from what look the unit looks.OB_SEG be as from the face of looking the polyhedron polymerization umbra (PAU) of looking the unit from looking actual the limit of the umbra boundary polygon (UBP) of looking the unit.

The border of blocking from looking the unit looks that runs in step 2015 is to be defined as uncontinuity grid (DM) the ambiguity line segment (DM_SEG) that blocks boundary segmentation (OB_SEG) of looking from looking the unit step 2063.The details of step 2063 is proposed subsequently.Step 2063 determines which DM ambiguity line segment is to be actually can run into from looking the border of blocking of looking the unit subsequently and to be performed in the past from block border and this step of looking the unit looks step 2015.

Mesh triangles shape and the visible cross-shaped portion of " about the wedge " of wedge represent umbra uncontinuity grid can or can be not with block segmentation corresponding to border (OB) from what look the unit looks.At step 2063(or not in step 2063) in block the border and store this result with the DM polyline what each DM polyline was defined as looking from the district.Because each DM polyline in step 2015 by it is processed by step 2063 before being run into, so the required information of the judgement in the step 2015 is determined in advance and stored for the DM ambiguity line segment that runs in step 2063.

If determining in decision steps 2015 that current triangle does not comprise blocks border (OB) segmentation, then treatment scheme marches to step 2035 to locate to interrupt current 3D trellis traversal in OB segmentation (OB SEG).Traversal can be passed leg-of-mutton other non-OB segmentations and be continued.Processing is interrupted at 2035 places.In other embodiment, the treatment scheme row marches to from 2035 and processes 20-1, and it is back to step 2025 with processing.Then, trellis traversal can be passed and is not that other mesh triangles shape limits of blocking boundary segmentation advance.

If determine that in decision steps 2015 current triangle does not comprise the OB segmentation, then treatment scheme marches to step 2020 to determine whether current triangle has silhouette edge.Should determine based on shown in Figure 3 for the test of single order from the silhouette edge looking the unit and look.

If determine that in decision steps 2020 current triangle does not have silhouette edge, then treatment scheme marches to step 2025, wherein marches to the triangle that the next one in the breadth first traversal of the digraph corresponding with flowing the shape triangular mesh does not travel through.

If determine that in decision steps 2020 triangle of processing comprises silhouette edge, then treatment scheme marches to step 2045, and wherein the breadth first traversal of the digraph corresponding with the triangular mesh single order method (Fig. 1 to Fig. 6) that is stopped and uses the rotation of wedge structure at the silhouette edge place and scan method is constructed a wedge that is incident on the silhouette edge.A wedge of constructing is added in the tabulation that is called as " wedge _ tabulation ".A wedge is to use rotation and scan those wedges that method is constructed at the single order silhouette edge that runs into.When initial construction, in some embodiments, all wedges are the initial wedges that does not also further limit by the observability step about wedge.Initial wedge structure in the step 2045 is corresponding at the initial wedge structure shown in the step 1210 of the output sensitive method of Figure 12.

Following alternate embodiments is possible: do not construct single order wedge (structure in step 2045 and subsequently step 2060) with the rotation of present embodiment and the method for scanning but alternatively use the not too accurate method that SV-ME wedge plane is extended to the intersection to construct some wedges (as being described in 1992 by people such as Teller, wherein only generating such plane on entrance sequence limit).Present embodiment comprises following technology: can be at inside corner summit place to by rotation with scan method and the difference of the umbra volume that the intersecting plane method produces is estimated.In this technology, (in Figure 20 B, it is described in detail), based on to the increase of desired precision and the observability figure that is blocked the result that boundary segmentation produces by additional SE-MV the complexity heuristics of carrying out balance select to be incident on for structure the method for the continuous conservative umbra event face on the inside corner summit.

Treatment scheme marches to step 2050 from step 2045, wherein selects the wedge in " wedge _ tabulation " and is committed to follow-up treatment step 2055 to 2068.

Treatment scheme marches to step 2055, and it is the first step that the wedge in " wedge _ tabulation " is processed, and wherein the visible segmentation about wedge of the mesh triangles shape of intersecting with wedge is determined.

This responsive 2D trellis traversal of effective output solution about the visibility problem of wedge has been proposed in Figure 15.The 2D that the method for Figure 15 is actually in the method shown in current Figure 20 implements, and in some embodiments, will be discontinuous grid segmentation (DM_SEG) about the visible fragmented storage of wedge.Should determine that method also identified the required compound-contoured summit (CSV) of subsequent step 2060 about the observability of wedge.

Treatment scheme marches to the CSV structure secondary wedge of step 2060 to identify at each wedge from step 2055 during previous step 2055.CSV be during step 2055 at wedge and single order from the visible cross-shaped portion about wedge of the silhouette edge looking the unit and look (such as identification in previous step 2055) generation CSV.Each CSV is the visible cross-shaped portion about wedge of wedge and single order silhouette edge.These put corresponding with the t joint of the compound profile contour of looking from the district.

In step 2060, according to some embodiments, each CSV structure SE-MV wedge of identifying.These wedges that come from the CSV are called as the secondary wedge.These wedges with the rotation of wedge structure and scan method scan that operation constructs.All these wedges that generate in this step all are to scan the SE-MV type that generates in the processing.This scans between the SVV of the original wedge of giving mutually the CSV place and carries out.Become continuous conservative umbra face by scanning the wedge of processing structure, its connection intersects at the original wedge at CSV place.In the embodiment that substitutes, can extend two former initial lines so that it intersects to form not too accurate but still conservative umbra border.

Limit such as the front, each CSV is corresponding to wedge and the cross-shaped portion that is supported on another wedge on the crossing silhouette edge.These wedges intersect at CSV point place.

If intersect in two SV-ME wedges rotation at CSV place to the identical summit, unit of looking, then two wedges accurately intersect at their place, limit and do not construct new SE-MV wedge.

To form by rotation to identical two summits looking cellular zone if intersect two wedges at the CSV place, then on the CSV to scan with rotary configured result be single SE-MV wedge.In this case, if two original wedges that intersect at the CSV place are SV-ME types, then be configured in CSV place connection SE-MV wedge conservatively approximate by looking element sides (connect two supports and look the summit, unit) and two the quadric surfaces that SV-ME silhouette edge form corresponding with the crossing wedge of CSV.Construct in this case the single SE-MV wedge quadric surface of the approximate correspondence that is formed by the S-EEE event conservatively on the CSV.The SE-MV triangle of in fact, constructing can be interpreted as having the quadric surface without the degeneration of string pitch.

Form to the summit that belongs to difference and look element sides if intersect two wedges at the CSV place and be by rotation, then the rotation on the CSV is the SE-MV wedge sequence that the limit connects with the result who is connected structure.In this case, if two original wedges that intersect at the CSV place are SV-ME types, the set of then constructing the connection SE-MV wedge on the CSV conservatively approximate since connect look the quadric surface that element sides and two other silhouette edges corresponding with the crossing SV-ME wedge of CSV form.Again, each SE-MV wedge can be thought the quadric surface of the corresponding degeneration with unlimited pitch.

Can also with the SE-MV wedge with generate the secondary wedge from CSV corresponding to the cross-shaped portion of looking the silhouette edge of looking the unit.

All the secondary wedges that to construct in step 2060 (being incident on those wedges on the CSV) are added into WEDGE_LIST, and it means that their can be finally process to find visible segmentation about wedge by step 2055.

Treatment scheme from step 2060 march to decision steps 2063 with determine DM_SEG from correspondence look the unit support the wedge that comprises DM_SEG support look cellular construction (summit or limit) every other " part " in addition and look and whether be blocked.The embodiment of this test is disclosed in Figure 21.Briefly, this test relates to and DM_SEG is intersected with all wedges that intersect potentially of having constructed and then tests to determine by the polyhedral some pardon of revising whether result's sub-segmentation is positioned at the PAU that looks from looking the unit.

If DM_SEG otherwise is blocked (except the SVV of the wedge that supports DM_SEG) from the every other part of looking the unit and looking, then DM_SEG is and blocks border (OB) from reality corresponding to the limit of looking the PAU that looks the unit from what look the unit looks.

Therefore, be from looking unit otherwise be blocked (the using the processing of Figure 21) of looking if in decision steps 2063, determine DM_SEG, then treatment scheme marches to 2068, wherein DM_SEG is added into crossing triangle as OB_SEG.

On the other hand, if DM_SEG otherwise is blocked from looking every other the part of looking the unit, then DM_SEG be not actual from look the unit looks block border (OB) and treatment scheme marches to step 2070 to determine whether remain any untreated wedge " wedge _ tabulation ".If determine the untreated wedge of residue in the wedge tabulation in decision steps 2070, then treatment scheme is back to step 2050.

On the other hand, if determine not remain untreated wedge in step 2070 in " wedge _ tabulation ", then treatment scheme marches to step 2080 to determine whether there is any untreated triangle in current " traversal " (being connected to current leg-of-mutton triangle set by non-silhouette edge).If determine to have the not triangle of traversal in decision steps 2080 in traversal (being connected to current leg-of-mutton triangle set by non-silhouette edge), then treatment scheme marches to step 2050, and the triangle that wherein next one is not traveled through is processed.

If in decision steps 2080, determine in traversal, not exist untreated triangle, then treatment scheme marches to step 2083, wherein those triangles that intersect with the OB segmentation that generates in current traversal is identified and is used crossing OB segmentation that those triangles are carried out again trigonometric ratio.This step can limit the 3D trellis traversal of blocking boundary that runs into subsequently effectively in step 2015, thereby traversal is restricted to the non-district that is blocked, and has further strengthened the output sensitive features of this method.

Treatment scheme marches to step 2085 and continues the 3D trellis traversal with the side that is not blocked in the OB segmentation that is generated by traversal.Treatment scheme stops at 2085 places.Because the wedge from 2-stream shape can produce the OB point different (unconnected) 2-stream shapes, so this is illustrated in the continuation of the traversal on the stream shape (part of PAU) of the new connection that the wedge line by OB section place forms by two " shelter fusions " of flowing shapes.

As previously mentioned, in a kind of embodiment of this method, the whole set of triangular mesh object is comprised by the mesh object of sealing.The mesh object of this sealing similar be generally used for around " day sylphon " for the mesh object of the visual application of interactivity.The sky sylphon that this grid traversal method among the 3D will seal is considered as flowing shape except any other the triangular mesh of looking the unit that is positioned at the sealing mesh object.Figure 20 K shows the mesh object of the sealing of such being labeled as " day sylphon ", and it comprises other polygonal mesh.The arrow that is labeled as " normal " represents: be different from other polygonal meshs, the sealing grid has all grid polygons that the normal that is oriented to them points to the inside of this grid partly.In conjunction with Figure 20 K other details of Figure 20 K have been carried out detailed discussion.

The method guarantees to travel through the omnidirectional's traversal that continues whole database, does not appear in the database even work as other mesh objects.The use of this sealing grid help so that current method for exporting sensitivity.That is, because only process the grid that is not blocked that runs into, so do not need each grid in the database is processed.In addition, this method can be identified the district that is not blocked of day sylphon self naturally, since size and the position of day sylphon, its source common or that obviously overdraw.

In a word, the general structure of the method that Figure 19, Figure 20 and relevant Figure 21 summarize is the breadth-first stream shape traversal structure that links up, it is interrupted by one of following three conditions usually: the mesh triangles shape (step 2010) of blocking potentially that 1) does not travel through, 2) in triangle, exist from what look the unit looks and block border (step 2015), or 3) there is single order silhouette edge (step 2020) at triangle.

Step 2010 guarantees that strict vertical stream shape is processed.Step 2015 prevents that with step 2085 district that in fact is blocked that travels through at grid from carrying out.Simultaneously, these steps just produce the output sensitive property at processed silhouette edge by what prevent many Grid Edges that are blocked and they.

The visible silhouette edge of step 2020 identification and beginning are following to be processed: generate a wedge and the secondary wedge (at the CSV place) that is associated and establish these wedges and mesh triangles shape between the visible segmentation about wedge of cross-shaped portion.These visible segmentations become conservative single order umbra uncontinuity grid segmentation, follow-up will the test it (step 2063 figure 21 illustrates the details of this test) with determine they whether be included in observability figure in from look the unit looks to block the border corresponding.

Similarity between 2D trellis traversal method and the 3D trellis traversal method is being obvious when with Figure 14 to Figure 19, Figure 15 to Figure 20 and Figure 16 to Figure 21 comparison.At Table VII (aspect the 3D traversal), Table VIII a(2D traversal aspect) and Table VIII b(1D traversal or ray projection aspect) in these similaritys also have been described in detail in detail.

Compare 3D trellis traversal aspect, 2D trellis traversal aspect and 1D trellis traversal (ray projection) aspect

Table VII

Be used for that the trellis traversal of looking the unit, conservative point are blocked (looking from looking the unit) inquiry and the aspect of the 3D trellis traversal method of the back projection of looking from silhouette edge from looking

Table VIII

Be used in that the observability about wedge, conservative point are blocked the inquiry of (looking from the limit) and the aspect of the 2D trellis traversal method of the back projection of looking from the profile summit

Table VIII b

The aspect of 1D trellis traversal (ray projection)

Figure 20 B illustrate use estimate at place, inside corner summit by rotation and scan method and the method for the estimated value of the difference of the umbra volume that the intersecting plane method produces, this difference is used for determining to construct at place, inside corner summit the method for continuous umbra event face.

According to some embodiments, when running into simple inside corner profile summit or compound-contoured summit (CSV), the processing shown in Figure 20 B is from step 2087.In embodiment, this occurs in step 2020 and 2055 places of Figure 20.

Treatment scheme marches to step 2089 to come tectonic line by the Plane intersects that makes the SV-ME wedge adjacent with the profile summit.Or in step 2089, being incident on the limit on the inside corner profile summit and determining SE-MV limit with intersecting plane formation minimum angles of structure SE-MV wedge.In one example, shown in Fig. 9 D, this angle is the line by line L1(intersecting plane) and the angle that produces of line L2, line L2 is the limit of corresponding SE-MV wedge.This angle is to measure at the cross-shaped portion at summit ICSV place according to these two lines.This minimum angles is stored in variable " DIFF_ angle ".

Treatment scheme marches to decision steps 2090 to determine whether angle " DIFF_ angle " surpasses predetermined value.

If determine " DIFF_ angle " greater than predetermined value in decision steps 2090, then treatment scheme marches to step 2092 and is incident on inside corner simply or the SE-MV event face on the compound-contoured summit with structure.Treatment scheme stops at 2092 places.

Be not more than predetermined value if in decision steps 2090, determine " DIFF_ angle ", then treatment scheme marches to step 2094, wherein constructs the continuous conservative umbra event face that is incident on the simple or compound-contoured summit of inside corner by the Plane intersects that makes adjacent S V-ME wedge.Treatment scheme stops at 2094 places.

Figure 20 C to Figure 20 J is polygonal mesh and the figure that looks the unit.These figure are used for explanation in the example of a kind of embodiment of the processing of the 3D trellis traversal shown in the process flow diagram of Figure 20 and relevant figure.

Figure 20 C to Figure 20 J shows from two polygonal meshs at two different visual angles and looks the figure of unit.In the upper diagram (Figure 20 C, Figure 20 E, Figure 20 G, Figure 20 I) of every figure, along observing said structure from polygonal mesh towards looking the unit.In bottom diagram (Figure 20 D, Figure 20 F, Figure 20 H, Figure 20 J), skeleton view is generally from looking the unit towards polygonal mesh.

The processing hypothesis of Figure 20 A selects initial delta to begin the 3D trellis traversal.Figure 20 C and Figure 20 D show the triangle that will be labeled as TB1 and are chosen as the leg-of-mutton example of beginning.TB1 is the grid polygon that is labeled as the polygonal mesh of " grid B ".In this example, embodiment adopts breadth first traversal, although also can adopt other traversal order.

The decision steps of Figure 20 A is determined by current triangle with look any whether the triangle that does not travel through is arranged in the 3D axle that the unit forms.Figure 20 E and Figure 20 F show and are formed on triangle T B1 and look 3D axle between the unit.This triangle-look unit shaft to be formed by eight triangles.In some embodiments, this axle is triangle and looks convex closure between the unit.Can adopt the convex closure building method of prior art to construct triangle and look axle between the unit, or any other protruding vision area.(referring to O ' Rourke, Computational Geometry in C Second edition Cambridge University Press1998, its full content merges among the application by reference).In one embodiment, reaching the rotation that relevant figure describes and the method for scanning in conjunction with Fig. 1 to Fig. 5 is suitable for constructing this axle by the SE-MV support polygon that structure additionally also is positioned on the external corners of grid.

Figure 20 E and Figure 20 F show axle TRI-VC-1 and intersect with other mesh triangles shapes that comprise the triangle T A1 that belongs to " grid A ".In this example, treatment scheme marches to the step 2030 of Figure 20 A.

In this example, the traversal of " grid B " is ended in step 2030 and is again initiated trellis traversal at the triangle T A-1 place of intersecting with axle TRI-VC-1.

After triangle T A1 place initiated breadth first traversal again, treatment scheme marched to step 2005, ran into not traversing triangle.

Treatment scheme marches to decision steps 2010, wherein in this example any untreated mesh triangles shape that is formed on triangle T A1 and look in the axle between the unit is detected.In this example, triangle T A1 and the axle of looking between the unit are empty, so that treatment scheme marches to decision steps 2015.

To determine whether triangle T A1 comprise any from looking the border of blocking that the unit looks for decision steps 2015 in this example.Because triangle T A1 does not comprise from looking the border of blocking that the unit looks, so treatment scheme marches to step 2020 to determine whether triangle T A1 has the silhouette edge that any single order is looked from looking the unit.

In this example, because triangle T A1 does not have any single order silhouette edge, so treatment scheme marches to the triangle that step 2025 does not travel through with the next one in the identification polygonal mesh.

The breadth first traversal of advancing from triangle T A1 in step 2025 in this example, runs into the triangle T A2 that the limit connects.In the step 2010 for triangle T A2 determined can be illustrated in TA2 and look triangle between the unit-look do not have the grid polygon in the unit shaft.

In this example, there is not the boundary edge of blocking that is incident on the triangle T A2 in step 2015 expression, so that treatment scheme marches to step 2020, determines that wherein triangle T A2 has the single order silhouette edge that is labeled as E2.

Treatment scheme marches to the step 2045 among Figure 20 A.In this example, the execution of step 2045 causes the wedge structure on the E2 of limit.Single order rotation (SV-ME) wedge that is arranged on the single order silhouette edge E2 of structure is marked as W2 at Figure 20 G and Figure 20 H in step 2045.Wedge W2 is rotation (SV-ME) wedge of constructing with the method that is used for structure as the rotation wedge described in Fig. 4 and relevant figure in step 2045.The support that is used for structure W2 look the summit, unit be in this example be labeled as SVV look the summit, unit.

In this example of the processing of wedge W2, treatment scheme advances to the step 2050 among Figure 20 A in causing step 2055.In step 2055, determine the visible segmentation on the wedge.In one embodiment, this determine to use as the 2D version of the trellis traversal method described in Figure 15 and relevant figure.In this example, the result who determines for the visible segmentation with respect to point (in this example with respect to SVV) on the wedge of wedge W2 is the uncontinuity grid segmentation (DM_SEG) that is labeled as OBE1.In this example, OBE1 be wedge W2 from the visible unique cross-shaped portion of SVV.

After having determined in this example DM-SEG OBE1, treatment scheme marches to step 2060, and step 2060 is to determine that a wedge W2 does not intersect with any single order silhouette edge in this example.So in this example, do not have compound-contoured summit (CSV) generation and do not need to construct the secondary wedge.

In this example, treatment scheme marches to decision steps 2063 to determine that as seen whether otherwise DM_SEG OBE1 look from looking the unit.This test determines that whether DM-SEG OBE1 is actual in looking the boundary segmentation that blocks that the unit looks.In one embodiment, this test utilizes the processing of Figure 21 A and relevant figure, and this processing comprises the 3D axle between the structure DM-SEG and determine whether to exist any not grid polygon of traversal in this axle, as represented in step 2110.In this example, OBE1 and the 3D axle looked between the unit are shown as mark axle SEG-VC-1.In one embodiment, follow the treatment scheme of process flow diagram Figure 24 A, Figure 24 B; In the example of OBE1, process any other wedge that can not identify among the axle SEG-VC-1, it can cut apart OBE1 potentially.In addition, in this example, the some test in the polyhedron of the Figure 25 that has made amendment with point on the DM_SEG OBE1; In this example, can determine that DM_SEG OBE1 is actually from looking the boundary segmentation that blocks that the unit looks, also therefore can be added to polygon " polygon 3 ", the initial polygon " polygon 3 " shown in Figure 20 C and Figure 20 D (being in this case quadrilateral) is divided into polygon " polygon 3R " again.

The processing that continuation is undertaken by the embodiment of Figure 20 A, to travel through all visible triangles of " grid A ", in this case, the set of the mesh triangles shape that has traveled through of " grid A " is corresponding to look all mesh triangles shapes of just facing on the summit, unit with respect at least one.Finally, breadth first traversal is ended at the first silhouettes place that comprises E2 of " grid A ", and it causes and blocks border OBE1 on " grid B ".In this example, single order silhouette edge E3(is shown in Figure 20 I and Figure 20 J) cause equally and block boundary edge OBE2(shown in Figure 20 J on " grid B ") and OBE3(shown in Figure 20 I and Figure 20 J).

These block boundary edge and comprise the continuous border of blocking, and it limits that it is divided at the OBE1 place again to " grid B " polygon 3(blocking boundary edge OBE2 and block OBE3 place, border subsequently) and other polygonal traversals of " grid B ".After trigonometric ratio step 2083 again, trellis traversal continues at " grid B ", and only continues in this side of not blocking of blocking the border, and is represented such as step 2085.

Figure 20 K shows and look the unit skeleton view of looking unit and polygonal mesh " grid A " and " grid B " identical with " grid B " with polygonal mesh " grid A " in Figure 20 C to Figure 20 J.In addition, Figure 20 K shows the closed polygon grid that is labeled as " day sylphon ".The observer polygon of the encirclement grid right with respect to the viewpoint back side is not shown so that can see the inside of the encirclement grid that is labeled as " day sylphon " in this skeleton view.The arrow that is labeled as " normal " represents: be different from other polygonal meshs, surround grid and have all grid polygons that the normal that is oriented so that them is pointed to the inside of grid partly.The wedge that does not have to intersect with other polygonal meshs can and seal Box junction at last.The same with any other wedge-grid intersection, utilization to wedge-grid intersection whether vision area block determining of border (step 2063), wedge-sealing grid intersection is processed according to coming according to the exemplary process diagram of Figure 20 A, and in the side of not blocking of blocking the border trellis traversal (step 2063) appears finally, here be labeled as " day sylphon " the sealing grid traversal appears on a side of inside.

Figure 21 A and Figure 21 B are for determining that the segmentation of uncontinuity grid is from looking the unit process flow diagram of method of (that is, whether the segmentation of uncontinuity grid was looked from the district and blocked the border) of whether being blocked of looking.

During main 3D trellis traversal shown in Figure 20 was processed, step 2063 was to determine to test from the judgement whether other " parts " 2DDM_SEG that looks the unit is blocked.Figure 21 shows a kind of embodiment of this test.

In some embodiments, if the part of not looking the unit from the DM segmentation except unity element is looked in the support that is used for the structure wedge corresponding with the DM segmentation is visible, then uncontinuity grid (DM) segmentation is from looking the border of blocking that the unit looks.In the method, this test can be organized as the back projection type observability inquiry of using the DM segmentation as line source and determining to look the observability of unit.

Figure 21 illustrates this test.

According to some embodiments, processing shown in Figure 21 originates in decision steps 2110 places, therein, has used the 3D axle between the DM_SEG that tests and has looked unit (" DM_SEG-looks the unit " axle).

If in decision steps 2110, determine in the DM_SEG-VC axle, to exist the not triangle of traversal, then treatment scheme advances to step 2115 and sentences abort process, and the major part that the 3D trellis traversal is processed jumps to not traversing triangle nearer in the DM_SEG-VC axle.This redirect has been guaranteed: determining that in fact whether DM_SEG is before look the rear-projection test of blocking border (OB) of looking the unit, on as identifying and process from all potential geometric configuratioies of blocking of looking the DM-SEG state that can affect that blocks the border of looking the unit.Treatment scheme stops at step 2115 place.

If determine in decision steps 2110 in the DM_SEG-VC axle that not the triangle of traversal does not exist, then treatment scheme marches to step 2120, and wherein DM_SEG intersects with other wedges in the DM_SEG-VC axle.This point of crossing can be divided into initial DM_SEG a plurality of DM_SUBSEG, and wherein each has the unified observability of looking from looking the unit.

Treatment scheme marches to decision steps 2125 to determine whether current DM_SEG or DM_SUBSEG directly link to each other with having determined DM_SEG or DM_SUBSEG from looking the observability state (crested or not crested) of looking the unit.

If determine that in decision steps 2125 current DM_SEG or DM_SUBSEG directly link to each other with having determined clearly DM_SEG or DM_SUBSEG from looking the occlusion state of looking the unit, then treatment scheme marches to step 2130, wherein the occlusion state of current DM_SEG or DM_SUBSEG is configured to and has knownly to looking the observability state of looking the unit and directly the state of the DM_SEG that links to each other of and current DM_SEG or DM_SUBSEG or DM_SUBSEG is identical, and this state is back to call function.Treatment scheme stops at step 2130 place.

If in decision steps 2125, determine current DM_SEG or DM_SUBSEG not directly with have known DM_SEG or DM_SUBSEG from looking the observability state of looking the unit and link to each other, then treatment scheme marches to step 2135 with the point on the selection DM_SUBSEG.

Treatment scheme marches to step 2140, the point of selecting in step 2135 forms axle between the unit and processing marches to step 2145 with looking, and wherein all wedges in the DM_SEG-VC axle is identified and place it in tabulation " wedge tabulation " with for subsequently processing.These wedges are the wedges that generate in step 2045 in Figure 20 and the step 2060, and main 3D trellis traversal is processed the single order wedge that has generated orthogonal projection.

Treatment scheme marches to step 2148, and wherein each wedge in " wedge tabulation " is processed by subsequent step.

Processing marches to step 2150, wherein for each wedge in " wedge tabulation ", determines this wedge whether single order wedge or high-order wedge.The main 3D trellis traversal of Figure 20 is processed with the rotation of the wedge at step 2045 and step 2060 place and is scanned structure and generates the single order wedge.Alternatively, can use in this instructions part subsequently and process to construct the high-order wedge in conjunction with Figure 22, Figure 23 and Figure 24 back projection described in detail.According to some embodiments, the rank of wedge are to determine according to the processing that is used for this wedge of structure.For example, the wedge that uses back projection to process structure is the high-order wedge, rather than the wedge that uses back projection to process structure is the single order wedge.If determine that in decision steps 2150 this wedge is the high-order wedge, then treatment scheme marches to step 2155.

On the other hand, if determine that in decision steps 2150 this wedge is not the high-order wedge, then treatment scheme directly marches to step 2158.

At step 2155 place, intersect by making high-order wedge and the every other wedge and the UBP that are arranged in wedge-look unit (" wedge-VC ") axle and point-look unit shaft, construct the actual umbra boundary polygon (UBP) corresponding with the high-order wedge of looking from looking the unit.For the high-order wedge of sign in step 2150, the part that is positioned at a little-looks unit shaft inside of corresponding UBP is constructed in this processing effectively.In some embodiments, can be by making corresponding wedge and intersecting to come according to this corresponding wedge structure UBP with the every other wedge/UBP that looks in the axle that forms between the unit at wedge.After step 2155, treatment scheme marches to step decision steps 2158 to determine that whether " wedge tabulation " be as empty.If " wedge tabulation " is empty, then treatment scheme is back to the step 2148 of the next wedge in " wedge tabulation " being carried out subsequent treatment.

If determine that in decision steps 2150 wedge is the single order wedge, then treatment scheme directly marches to aforesaid step decision steps 2158.

If it be empty determining " wedge tabulation " in decision steps 2158, then treatment scheme marches to step 2160, the point in the polyhedron of wherein the DM_SUBSEG point of selection in step 2135 being made amendment test with definite its whether from looking the unit crested of looking.

Figure 25 illustrates to the embodiment of the polyhedral some pardon test of revising and in conjunction with this figure and be described in detail.Briefly, polyhedral some pardon test of this modification determines whether this test point is positioned at from looking the inside of the PAU that looks the unit.Look on the elemental area in the situation that point is positioned at, this test and traditional polyhedral some pardon test select aspect the known second point that is positioned at the PAU outside a bit similar.Being configured with line segment and this line and polyhedral polygon between this test point and this external point intersects.The state of the sub-segmentation of this line (this polygon inner or outside) will be based on the segmentation of grid polygon with comprise that the cross-shaped portion of the UBP of PAU changes.In the situation that the single order wedge, needn't intersect with other wedges at the single order wedge can not be crossing with other wedges in their side of not blocking because of these single order wedges with the corresponding UBP(that forms).In addition because only be positioned at these PAU faces that test point looks unit shaft just can with the test line segment intersection, finish test so needn't construct whole PAU.

Follow step 2160, treatment scheme marches to processes 21-1, and it starts from step 2165 place (Figure 21 B).

In decision steps 2165, the result who the polyhedral some pardon of revising is tested (step 2160) checks.

If determine the test point crested in decision steps 2165, then treatment scheme marches to step 2175, and wherein overall process is returned the result that the DM_SEG of test is blocked.This result is back in the step 2063 among Figure 20.

If determine that in decision steps 2165 test point does not have crested, then treatment scheme marches to step 2170, and wherein overall process is returned the result that the DM_SEG of test is not blocked.This result is back to step 2063 among Figure 20.Treatment scheme stops in step 2170 and step 2175 place.

The observability figure that the conservative single order of the method construct of Figure 19, Figure 20 and Figure 21 is looked from looking the unit.The method has been utilized the conservative linearization umbra event face (CLUES) of on single order silhouette edge incident, and determines that in fact whether corresponding uncontinuity grid segmentation is to block border (segmentation of PAU) from what look the unit looks.

In the method, according to some embodiments, the syndeton on umbra DM border is used for accelerating significantly determining that whether the DM border is to block border (limit of corresponding VM) from what look the unit looks.The observability state of structure (for example, DM polyline) (from look the unit look be blocked or be not blocked) can be only when this structure and visibility events face intersect, just change.For the DM polyline, this occurs over just in the situation that this DM polyline and another DM polyline intersect.Between these intersected, the observability state of DM polyline did not change.Thereby, for the observability state of a plurality of segmentations between cross-shaped portion of determining the DM polyline, only need to determine the state of a segmentation.For the adjacent sectional between the cross-shaped portion of DM polyline, only generate the event face that blocks potentially/expose.This is by at each DM ambiguity line segment and look axle between the unit to determine that whether having any untreated geometric figure that blocks potentially in this axle guarantees.The same with all the axle tests in the present embodiment, the simple hierarchical space tissue of geometric database can be guaranteed O(nLogn) performance.

Generally, the responsive structure of the output of conservative linearization umbra observability figure adopts following methods: the traversal that it has used the sudden change profile contour from polygonal mesh 2 dimension stream shapes (being embodied in the 3D) to stream shape and has blocked the border.This travels through the complete mode that flow to realize to export responsive observability cascade of effectively cascade and visibility events face (the profile contour on the stream shape of other that wedge/UBP) is crossing.

Figure 21 C shows according to strong visible, unshielding or is just facing all the time the PVS polygon is classified.

According to some embodiments, PVS is comprised of the visible polygon segmentation of conservative observability figure, and this conservative observability figure generates by the 3D traversal method of Figure 20 and correlogram and the alternate embodiments with UBP of Figure 26.Alternatively, derive PVS from the discontinuous grid method of the umbra of Figure 12.

No matter use in the embodiment which, concerning and classify easily from looking observability that PVS polygon segmentation that the observability of looking the unit generates all is based on them and looks the unit with the object-space of this method.

In some embodiments, the processing shown in Figure 21 start from decision steps 2180 with the polygon of determining observability figure whether from look the unit look into strong as seen.In some embodiments, will from look the unit look for strong visible polygon definition be following polygon: at this polygon and look 3D axle between the unit and do not comprise other polygon.In some embodiments, this determines to make during the observability figure for each polygon segmentation constructs processing (Figure 20).Thereby, if, during structure VM, can and be labeled as with strong visual identification (step 2180) " not having z relatively " as identifying as shown in the step 2183.If triangle is not that by force as seen then treatment scheme marches to step 2184, does not wherein have the mark triangle.

Because just can not blocked by any other polygon as long as viewpoint is arranged in the polygon of looking any quilt in unit mark like this of structure VM/PVS, so can play up the polygon of any quilt mark like this subsequently, and need in polygonal Z buffer zone for this, not calculate the z compare operation.

In follow-up judgement/classifying step 2185, determine whether this polygon among the VM is arranged in any other polygonal polygon of VM-look unit shaft.If current polygon is not arranged in any other polygonal polygon of VM-look unit shaft, then processes and march to the step 2187 that polygon is labeled as " not having z to write ".Otherwise treatment scheme marches to does not have the leg-of-mutton step 2186 of mark.

The polygon that is not arranged in any other polygonal polygon of VM-look unit shaft can not block any polygon of VM/PVS, so in the situation that do not have z relatively or z write this polygon played up, this can reduce to play up this polygonal cost significantly.

Finally, can by determine polygonal plane whether with look judgement/classifying step 2190 of intersecting the unit with each polygon classification among the VM/PVS.If in decision steps 2190, determine this polygonal plane not with look the unit and intersect, then process marching to the step 2190 that the VM/PVS polygon is labeled as " without back side rejecting ".

It all can not be the back side that these polygons are looked from any viewpoint of looking the unit that is used for structure VM/PVS.Thereby, do not need the back side to reject during the polygon that has this mark in to PVS is played up, thereby reduced to play up polygonal cost.If the test of step 2190 determines leg-of-mutton plane and look the unit and intersect, then process marching to step 2190 and this triangle of mark not.

Figure 22 shows the process flow diagram of processing for the control of 3D trellis traversal method, this 3D trellis traversal is used for the observability figure structure back projection, that look from silhouette edge, and this figure is used for determining looking summit, unit (VSVV) and unit profile contour (VSVSC) is looked in visible support from the visible support that silhouette edge is looked.

The preceding method of Figure 19, Figure 20 and Figure 21 only uses at the single order of single order incident from the silhouette edge of looking the unit and looking, conservative, linearization umbra event face (CLUES).

In the method for Figure 22, Figure 23 and Figure 24, single order CLUES be " tuned " for reflect more exactly high-order from looking the observability cross-shaped portion of looking the unit.

In some embodiments, this method has used the conservative property back-projection approaches that focuses on to look summit, unit (VSVV) to identify visible support of the conservative property (looking from silhouette edge) of constructing the high-order wedge thereon.In some embodiments, wedge should " adjusting " only be applied in following situation: be used in support in the structure of single order wedge of initial forward projection and look summit, unit (SVV) and blocked by the silhouette edge that is supported.Otherwise single order event face (wedge) is accurate.

Usually, only this adjusting is applied to coarse those parts of single order wedge.Coarse those parts of single order SV-ME wedge are corresponding with supported those parts of blocking on summit, unit (SVV) of looking of the silhouette edge that supports.The silhouette edge and the support triangle between the SVV that are formed on this wedge by inspection describe definite these parts of changing places.

Similarly, the out of true of single order SE-MV wedge part is corresponding to these parts of wedge: support and look unit profile contour (SVSC) and look from the grid profile summit of supporting and be blocked.Also by being scanned triangle inspection, the grid profile summit of wedge and the support between the SVSC segmentation determine this part.

Usually, in this method of Figure 22, Figure 23 and Figure 24, by adjusting to single order SV-ME wedge is determined as the back-projection approaches of light source in the institute support profiles limit of wedge, and use the 3D trellis traversal observability Cascading Methods of looking from the district and construct the observability figure that looks from the limit that guards that looks on the elemental area.This observability figure can comprise that (looking from the limit) visible support look summit, unit (VSVV), substitutes SVV with VSSVV and supports corresponding high-order wedge.Can have corresponding visible support depending on each the back projection observability figure that looks from the limit on the elemental area and look unit profile contour (VSVSC).

Similarly, in the single order observability, can connect adjacent SV-ME wedge via the SE-MV wedge that support triangle shape forms that scans of looking by extension unit supports looks unit contour (SVC) supports.For " adjusting " is incident on single order SE-MV wedge on the grid profile summit, also determine that by back projection visible the support look unit profile contour (VSVSC): structure from silhouette edge look about looking the observability figure of unit.This backprojection uses the silhouette edge of sharing as the grid profile summit of light source.Describe in detail such as the further part at this instructions, the structure that connects the high-order SE-MV wedge of two adjacent high-orders or " adjusting " SV-ME wedge depends on the relation of two VSVCSs corresponding with the SV-ME wedge, as determined in back projection is processed.

Therefore, the purpose of the backprojection that looks from the limit of Figure 22, Figure 23 and Figure 24 be structure from silhouette edge look about looking the observability figure of unit, according to its corresponding high-order wedge of deriving.In some embodiments, this conservative observability figure uses the modification of the 3D trellis traversal method of the observability cascade of having discussed in conjunction with Figure 19, Figure 20 and Figure 21 to construct.Although the method for Figure 19, Figure 20 and Figure 21 is used and looked the unit as the observability source, the method for Figure 22, Figure 23 and Figure 24 uses concrete single order silhouette edge (or the segmentation on limit) to solve visibility problem as " source " or vision area.Because the two is the visibility problem of looking from the district, so between the processing (and figure of correspondence) of correspondence, have obvious similarity.

From the important difference between the back projection observability method of looking the orthogonal projection observability solution of looking the unit and looking from the limit be: although the orthogonal projection method must solve all observabilities of looking from looking the unit, backprojection only determines the observability of looking from silhouette edge of the observability of looking from the limit of video display unit potentially.The back projection problem of therefore, looking from silhouette edge is restricted to and is formed on silhouette edge and looks axle between the unit.

3D trellis traversal for back projection: only adopt that orthogonal projection-leg-of-mutton method is not blocked

In a kind of enforcement of the method, in back projection, only use visible those triangle/segmentations orthogonal projection (unit looks from looking).In the method, visible triangle calculates their planar orientation of counter-rotating with respect to back projection in orthogonal projection.Usually, the orthogonal projection silhouette edge is identical limit with back projection silhouette edge (when from looking as the grid silhouette edge of putting for " source " of back projection).In this case, the triangle that in fact before application-oriented counter-rotating, does not have orthogonal projection towards rear-projection eikonogen limit (single order silhouette edge)-be not blocked.

In other cases, the visible grid elements of not every orthogonal projection can be with respect to the back side, limit, re-projection source pair.In these cases, if leg-of-mutton orientation is inverted, then it can become with respect to the back side, rear-projection eikonogen limit pair.This meeting so that its with respect to the back side, source pair, thereby create grid with the triangle with inconsistent orientation.In this case, the back side that will develop with respect to back projection point removes from grid the triangle that is orientated.

In back projection, fail to comprise the precision of the relevant adjusting of VSVV that the geometric figure that blocks potentially can reduce to calculate and wedge.But still guarantee the result that guards, this is because any geometric elimination of blocking potentially only can be so that the VSVV that calculates be the more conservative estimation of SVV.In addition, by only adopting back projection-visible triangle, reduced significantly the cost that back projection is processed, this is because initiating to have determined relevant orthogonal projection-visible triangle when back projection is processed for specific grid silhouette edge.

In the method, the visible grid elements of orthogonal projection is in fact as the virtual shelter that calculates back projection.

Therefore, select seed 3D grid in step 2205, this selection is made from the set of visible orthogonal projection (not being blocked) grid.

3D trellis traversal for back projection: the retraction of the silhouette edge under the high-order observability

In a kind of embodiment of this method, only according to determining back projection as those rear-projection eikonogens (being the grid silhouette edge in this case) on the single order profile summit the orthogonal projection (when when looking the unit and observe).In this is implemented, only those wedges that are attached to single order orthogonal projection profile summit are carried out " adjusting " with the observability of reflection high-order about the plane by back projection.The wedge of regulating keeps being attached to the original mesh silhouette edge.

In fact, when any limit of observing from VSVV as single order orthogonal projection silhouette edge (when observing from SVV), owing to sharing two triangles on this limit with respect to the VSVV back side pair, so this limit no longer is silhouette edge.Under these circumstances, orthogonal projection silhouette edge (rear-projection eikonogen) is in fact looked and is blocked from looking the unit.In this case, actual (accurately) the umbra event face of looking from the district is not to be supported by this orthogonal projection silhouette edge but look nearer limit, unit and support by being positioned at the identical upper distance of 2-stream shape.Single order be potential effect such as the high-order observability of being determined by back projection from what look the unit silhouette edge looks towards this migration of looking the unit or " retraction ".Fail to explain the precision that the correspondence of VSVV that this profile migration may reduce to calculate in some cases and wedge is regulated.But, always can produce the more conservative umbra event face of corresponding event face that is attached to the retraction profile of comparing because be attached to the visibility events face (being wedge in this case) of the adjusting of orthogonal projection single order silhouette edge.

Be attached to the wedge of the adjusting of its original single order orthogonal projection silhouette edge by maintenance, " the support segmentation " between silhouette edge and VSVV of the wedge of adjusting can be violated potentially by penetrating grid surface " the local observability " of the grid at its place.Under these circumstances, accurately, silhouette edge can be inclined to the unit moves so that the nearer limit that is connected of grid produces the wedge of the adjusting of not violating local observability towards looking, and such adjusting wedge becomes " accurately " silhouette edge.

In order to keep being attached to the conservative adjusting wedge of its original single order orthogonal projection silhouette edge, the back projection Treatment Design is become by regulating wedge to prevent from violating local observability.This occurs in following situation: regulate wedge and come from the orthogonal projection single order silhouette edge so that wedge is positioned at the inside of mesh triangles shape 2-stream shape.This is can be by flowing shape from type of barrier to forming the 2-that all grid elements (being triangle in this case) of the assembly is connected prevent with the orthogonal projection silhouette edge, and the assembly that wherein connects is fully in the axle of looking between unit and the limit.In some embodiments, can travel through to carry out with the simple triangular mesh that starts from the limit and when intersect on triangle and axle border, finish.These grid elements are those grid elements of certainly blocking that can produce the type of the retraction that causes silhouette edge.By determining to identify according to back projection and eliminating these elements, back projection can produce conservative " adjusting " wedge that keeps being attached to original orthogonal projection limit, but its explanation is so that SVV becomes the high-order observability effect of looking and being blocked from silhouette edge.

Therefore, the step 2205 of Figure 22 selects to be used for the seed 3D grid elements of back projection, does not consider those grid elements of certainly blocking that these can cause back projection-source silhouette edge potentially.

The method also simplified use the 2D trellis traversal about the enforcement in the observability situation of wedge, as discussing in conjunction with Figure 17.When being used in, the method use realized even larger simplification this situation lower time from looking the observability of looking the unit of 3D trellis traversal.By adopt the method in 3D trellis traversal situation, event face (wedge polygon) keeps being attached to the umbra boundary surface of from district looking of original single order orthogonal projection profile contour to guarantee to construct continuously, to guard.

Again, the embodiment that substitutes is possible: by explaining from blocking the in proper order retraction of silhouette edge.For example, the method for a kind of searching " retraction " silhouette edge is calculated back projection according to initial projection, single order silhouette edge and is determined whether the wedge corresponding with the VSVV that calculates violates local observability.If violate local observability, then in fact candidate's silhouette edge is sightless from looking the unit looks when considering the high-order effect, and apart from look the nearer adjacent edge in unit then as rear-projection eikonogen to determine whether wedge (corresponding with the VSVV on new candidate limit) violates local observability.Can repeat this processing until run into the wedge of not violating local observability.With this wedge as regulate, the high-order wedge.

Usually, although the complexity of the increase that back projection is processed can be clearly, only normally very little by the additional amount of blocking of looking from the district of the retraction generation of silhouette edge.Remaining fixed to the orthogonal projection wedge of original orthogonal projection single order silhouette edge and calculating " adjusting " wedge that is incident on the same profile limit (corresponding with VSVV) by back projection is a kind of method of having explained most of high-order observability effects with very little increase cost.

Emphasize such as the front, the high-order back projection is regulated only to be applied in when original SVV looks from single order silhouette edge or its segmentation and is blocked.In some other embodiments, adopt to measure the maximum possible deviation between the high-order wedge of determining single order wedge and corresponding adjusting.For example, be maximum possible angular deviation between single order wedge and the high-order wedge by single order wedge (being included on the supporting plane) and the angle that forms at silhouette edge and the corresponding separating plane of looking between the unit; Because use this method, so the high-order wedge can be positioned at (if VSVV is burble point) on the separating plane under limiting case.This angle is tended to naturally along with reducing (vision area seems more " pointed " from rear-projection eikonogen) apart from the increase of the distance of looking the unit.Thereby the high-order observability effect of determining by back-projection approaches becomes not too important at a distance in the distance vision area.This angle tolerance can be used in and determines whether and calculate the high-order effect for given orthogonal projection single order silhouette edge.

Now return Figure 22, (step 2215) processed in the control of the major part of calling back projection 3D traversal discussed.Except with control from look the 3D trellis traversal method (Figure 19) of looking the unit similar, method shown in Figure 22 is also closely similar with look from the summit method of back projection of (2D trellis traversal) of previous described control shown in Figure 14.

Process flow diagram shown in Figure 22 shows initiation, again initiates and stop the processing of the back projection 3D traversal processing that principal and subordinate's silhouette edge looks, and shows in detail this processing and at the further part of this instructions it is described in detail in Figure 23.

In some embodiments, processing shown in Figure 22 starts from 2205, wherein selects " seed " triangle to process from the 3D trellis traversal that silhouette edge is looked initiating for single silhouette edge, as shown in the step 2205.When being in the orthogonal projection processing, select to be likely that the initial seed triangle that exposes shelter accelerates this processing.

In step 2205, only be only kind of a sub-grid candidate at silhouette edge and those triangular mesh of looking in the axle (SILE-VC axle) between the unit.

Treatment scheme marches to step 2210 is identified 3D kind sub-grid with the cross-shaped portion of the boundary polygon by limit-look unit shaft and seed mesh triangles shape containing segmentation.

When having identified seed 3D grid and plant containing segmentation on the sub-grid, then treatment scheme marches to step 2215 and processes to call the main of back projection 3D trellis traversal.Figure 23 illustrates the embodiment of processing for the main 3D trellis traversal of back projection.

Treatment scheme marches to decision steps 2220 to determine whether to exist the triangle of the major part of having ended the 3D trellis traversal.The 3D mesh triangles shape of if there is no ending, then treatment scheme marches to step 2225, and wherein the 3D trellis traversal is terminated, thereby has finished 3D about the structure of the observability figure that looks the unit.

Ended the 3D mesh triangles shape (in the step 2330 of Figure 23) of main 3D trellis traversal if determine existence in decision steps 2220, then treatment scheme marches to decision steps 2230 to determine whether there is any triangle in the axle that is formed by the triangle (TRI) of ending and source silhouette edge (SILE).

If determining in decision steps 2230 does not have triangle in the axle between current triangle and source silhouette edge (SILE), then treatment scheme marches to step 2255 to contain segmentation by being connected to the 3D triangular mesh structure that the leg-of-mutton triangle of current termination forms.

Treatment scheme marches to step 2260 and again initiates the 3D trellis traversal with the triangle to previous termination.

If determine to have triangle in decision steps 2230 in by the axle between current triangle and the source silhouette edge, then treatment scheme marches to step 2235 to determine whether any not triangle of traversal is arranged in the TRI-SILE axle.

If determine in the TRI_SILE axle to exist the not triangle of traversal in decision steps 2235, then treatment scheme marches to step 2240, wherein on the triangle of the nearer not traversal in the TRI-SILE axle with 3D trellis traversal " redirect " or again initiate.This step (it also occurs in the step 2330 of back projection master 3D traversal processing shown in Figure 23) was guaranteed before traversal continues and was handled in the past the triangle that all block potentially making the decision that stops traversal (fully being blocked because the triangle of termination is looked from the limit, source).

If determine not exist not the triangle of traversal in the TRI-SILE axle in decision steps 2235, then treatment scheme marches to step 2245 and represents from the look any polyhedron polymerization umbra (PAU) of one or more polyhedron volumes of being blocked of source silhouette edge to determine whether point on current (terminations) triangle is positioned at.In some embodiments, this test can be adopted polyhedral some pardon test of modification shown in Figure 25.Single order event face (single order wedge) is used in polyhedral some pardon test of this modification; And alternatively, high-order umbra boundary polygon (UBP).This test allows needn't construct whole polyhedron/polygon for the implicit solution of polyhedral some pardon test.

If the point of determine ending in decision steps 2245 on the triangle is to be positioned at PAU, then treatment scheme is advanced step 2250 again initiating traversal to ending triangle, and no longer will end triangular marker and be termination.

If determine that in decision steps 2245 ending triangle is not positioned at any PAU, then treatment scheme marches to step 2255, wherein constructs as previously mentioned the containing segmentation of the 3D grid that is associated and again initiates traversal.

As shown in Figure 22, treatment scheme stops at step 2225,2240,2250 and 2260 places.

Figure 23 shows the process flow diagram that is configured to determine look from the visible support that silhouette edge is looked main the processing of summit, unit (VSVV) and the visible 3D trellis traversal method that supports the observability figure that the back projection of looking unit profile contour (VSVSC) looks from silhouette edge.

The back projection 3D trellis traversal observability Cascading Methods of looking from silhouette edge main processed with some is similar from looking the 3D trellis traversal processing (Figure 20) of looking the unit.

Turn to now Figure 23, in some embodiments, main back projection 3D trellis traversal process with as filling traversal with the breadth-first of stream shape triangle or other polygonal meshs or flood and begin shown in the initial step 2305.In processing, the control shown in the process flow diagram of Figure 22 uses the initial delta of initiating whole processing.

The back projection of looking from silhouette edge can be processed and be applied to whole single order silhouette edge.In addition, can segment each single order silhouette edge (SILE) and calculate back projection for every sub-segmentation.Adopt the method in the adaptive refinement of each silhouette edge, wherein the single conservative wedge of refinement converges on the conservative linearization visibility events face of the accurate visibility events face (normally quadric surface) that is incident on the silhouette edge with generation.In Figure 22, Figure 23 and Figure 24, silhouette edge (SILE) can refer to single single order silhouette edge or its sub-segmentation.

As previously mentioned, in one embodiment, grid is stream shape triangle, and wherein each limit is accurately shared by two triangles.Other restrictions of grid also are possible, comprise inc grid.The cross-shaped portion of mesh triangles shape and wedge or UBP can produce more complicated grid polygon.Can use the polygon Triangulation Method polygon that these are more complicated of establishing to be transformed into triangle.Triangular mesh is expressed as digraph.The method also is fit to the grid polygon except being fit to triangle, but uses triangle to simplify this traversal method.The method of Figure 23 has adopted every stream shape triangular mesh that the limit is accurately shared by two triangles.

Step 2305 is indicated to the traversal of mesh triangles shape.Can travel through with breadth-first or flood fill order.Use the triangular mesh traversal of breadth-first, travel through from initial delta towards external expansion to form the layer of institute's traversing triangle.The triangle that connects by non-silhouette edge is said to be " traversal " that belongs to identical.

Other traversal order are possible.Select not affect for the triangle of the traversal of initiating one or more polygonal meshs the output of traversal.But, by selecting fully visible and near the initial delta of looking the unit, processing being easy to carry out better.In the processing that keeps the digraph traversal, step 2305 expression traversal is limited to the not traversing triangle in the grid.

Treatment scheme marches to decision steps 2310 to determine being formed by current triangle and source silhouette edge whether any not triangle of traversal is arranged in the 3D axle (this axle is called as the TRI-SILE axle) from step 2305.If one or more not triangles of traversal are arranged in this axle, then the treatment scheme row proceeds to step 2330, and wherein traversal jumps to the nearer not triangle of traversal immediately.In one embodiment, organize the grid polygon with the hierarchical space sub-structure.This is so that the polygonal large grouping of grid that the front travels through can be promptly identified in the axle test of step 2310.

In fact step 2330 carries out strict vertical grid elements processing for working as a grid or a plurality of mesh object.Traversal is jumped to the nearer leg-of-mutton processing of blocking potentially can produce the endless circulation that causes because leg-of-mutton circulation is overlapping.In Weiler-Atherton observability algorithm, also run into such circulation and can represent by the digraph of safeguarding overlapping relationship to come such looping detected and eliminate.In Figure 13, propose and at length the embodiment of the method is discussed in conjunction with this figure.

Step 2330 is guaranteed before current leg-of-mutton traversal the event face that comes from the nearer not traversing triangle in current leg-of-mutton triangle X silhouette edge (TRI-SILE) axle to be identified and constructed.These event faces can be to block border (determining for these nearer silhouette edges to 2368 in step 2345) from what look the unit looks, and in fact it can limit current leg-of-mutton traversal (step 2335).Jump to nearer not traversing triangle/silhouette edge by forcing, step 2310 guaranteed before current traversing triangle that structure can limit current traversing triangle anyly blocks the border from what silhouette edge was looked.Treatment scheme stops at step 2330 place.

If the triangle of traversal not in the TRI-SILE axle, then whether treatment scheme marches to decision steps 2315 and comprises from silhouette edge the blocking the border or comprising border (comprising the expression SILE-VC) of (" from-SILE ") of looking with the current triangle that is traveling through.Umbra uncontinuity grid (DM) segmentation generally forms polyline in the intersection of mesh triangles shape and wedge.Each DM segmentation is wedge and leg-of-mutton cross-shaped portion, wherein this segmentation from the support source element of wedge (point or limit) look and the outline elements that supports (point or limit) by wedge as seen.This is called as the visible cross-shaped portion of " about wedge ".This cross-shaped portion is the segmentation of the umbra uncontinuity grid guarded and can or can is not polyhedron polymerization umbra (PAU) limit.These visible segmentations about wedge comprise the polyline of conservative linearization umbra uncontinuity grid.These polylines can be or can not be actual look from the source (looking from silhouette edge in this case) block the border (above-mentioned border will be from the source (SILE) the conservatively visible polygon segmentation of looking separate with the polygon segmentation that is blocked of looking from the source, wherein the size of this life volume is conservatively underestimated).

The border of blocking of looking from the source that runs in step 2315 is to be confirmed as uncontinuity grid (DM) the ambiguity line segment that blocks the border of looking from looking the unit step 2363.Provide subsequently the details of step 2363.Step 2363 determines which DM ambiguity line segment is to block the border from what silhouette edge was looked, and in fact in later step 2315, can run into from look the unit looks block the border before carry out step 2363.

The visible cross-shaped portion of " about the wedge " of mesh triangles shape and wedge represents the segmentation of umbra uncontinuity grid, its can or can be not corresponding with look from silhouette edge block border (OB).What in step 2363 each DM polyline is defined as looking from the district blocks border (or uncertain block border for what look from the district) and is stored together with the DM polyline.Because process each DM polyline by step 2363 before in step 2315, running into each DM polyline, in step 2363, be determined in advance and be stored for the DM ambiguity line segment that runs into so be used for the information of the judgement of step 2315.

If determining in decision steps 2315 that current triangle comprises blocks border (OB) segmentation or comprise boundary segmentation, then treatment scheme marches to step 2335 with in OB segmentation (OB SEG) or block boundary and interrupt current 3D trellis traversal.Traversal can be passed leg-of-mutton other non-OB segmentations and be continued.Treatment scheme marches to step 2335.In other embodiment, treatment scheme marches to from 2335 and processes 23-1, and it is back to step 2325 with treatment scheme.

If determine that in decision steps 2315 current triangle does not comprise the OB segmentation, then treatment scheme marches to step 2320 to determine whether current triangle has silhouette edge.In some embodiments, this determines the test based on silhouette edge of looking from the limit for single order shown in Figure 3.

If determine that in decision steps 2320 current triangle does not have silhouette edge, then treatment scheme marches to the triangle that step 2325 does not travel through with the next one in the breadth first traversal of processing the digraph corresponding with stream shape triangular mesh.

If determine that in decision steps 2320 triangle of processing comprises silhouette edge, then process and march to step 2345, wherein the breadth first traversal of the digraph corresponding with triangular mesh stops at the silhouette edge place, and the single order method (Fig. 1 to Fig. 6) of using the rotation of wedge structure and scanning method is constructed a wedge that is incident on the silhouette edge.In embodiment, a wedge of constructing is added in the tabulation that is called " wedge _ tabulation ".A wedge is to use rotation and scan method construct those wedges on the single order silhouette edge that runs into.When initial construction, all wedges all are the initial wedges that also is not subjected to about the further restriction of observability step of wedge.In some embodiments, the initial wedge structure (exporting insensitive method) shown in the step 1210 among the structure of the initial wedge in the step 2345 and Figure 12 is corresponding.

Following alternate embodiments is possible: do not use rotation of the present invention and scan method but alternatively use to extend SV-ME wedge plane to the not too accurate method (as described in 1992 by people such as Teller) that intersects and construct single order wedge (structure in step 2345 and later step 2360).

Treatment scheme is advanced and is marched to step 2350 from step 2345, wherein selects the wedge in " wedge _ tabulation " and is submitted to treatment step 2355 to 2368.

In the first step to the processing of the wedge in " wedge _ tabulation ", step 2355 is determined the visible segmentation about wedge of the mesh triangles shape of intersecting with wedge.In some embodiments, this determines to use following the processing: all mesh triangles shapes and each wedge are intersected, determine that with Weiler-Atherton 2D is about the observability of wedge subsequently.The insensitive conservative umbra uncontinuity grid method of better simply output that the method is used in the uncontinuity grid method of prior art and is used in Figure 12.Alternately, in Figure 15, provided for this responsive 2D trellis traversal solution of more effective output about the visibility problem of wedge.Treatment scheme marches to step 2360 from step 2355, and the compound-contoured summit (CSV) on the wedge of being incident on of wherein processing is identified as wedge and single order from any observability cross-shaped portion about wedge of the silhouette edge looking the unit and look.

Each CSV is the visible cross-shaped portion about wedge of wedge and single order silhouette edge.These are corresponding to the t joint of the compound profile contour of looking from the district of correspondence.The compound-contoured contour is not by can being that silhouette edge by the summit on the primary flow shape directly links to each other forms.

In this external step 2360, each CSV structure SE-MV wedge of identifying.These wedges that come from CSV are called the secondary wedge.Construct these wedges with the rotation of wedge structure and the operation of scanning of scanning method.In some embodiments, all these wedges that generate in this step are to scan the SE-MV type that generates in the processing.The umbra face that these wedges become continuously, guard, it is connected to the SV-ME wedge that intersect at corresponding CSV place.Alternately, load remains conservative umbra border to form not too accurately to intersecting can to extend two SV-ME wedges.

The alternate embodiments that employing is constructed to the wedge of the method that intersects by extension adjacent S V-ME wedge is possible.In this enforcement, do not construct the SE-MV wedge.

Defined such as the front, each CSV is corresponding to wedge and the cross-shaped portion that is supported on another wedge on the crossing silhouette edge.These wedges intersect at CSV point place.

In addition, if two SV-ME wedges that intersect at the CSV place rotate to identical silhouette edge summit, then two wedges locate accurately to intersect and do not construct new SE-MV wedge on their limit.

All the secondary wedges (being incident on the CSV) that to construct in step 2360 are added in " wedge _ tabulation ".It means that their finally can process to find visible segmentation about wedge by step 2355.

Whether treatment scheme marches to decision steps 2363 from step 2360 and is blocked to determine that DM_SEG looks from every other " part " except support comprises the summit of wedge of DM_SEG of the silhouette edge (SILE) of correspondence.The embodiment of this test is disclosed in Figure 21.Briefly, this test relates to intersects DM_SEG and all and all wedges that intersect potentially of having constructed, and then tests to determine by the polyhedral some pardon of revising whether result's sub-segmentation is positioned at from the source PAU of look (looking from silhouette edge in this case).

(except the SVV of the wedge that supports DM_SEG) otherwise is blocked if DM_SEG is the every other part of looking from silhouette edge, then DM_SEG be the actual PAU with looking from the limit corresponding block border (OB) from what silhouette edge was looked.

Therefore, be otherwise be blocked (the using the processing of Figure 21) of looking from SILE if in decision steps 2363, determine DM_SEG, then treatment scheme marches to 2368, wherein DM_SEG is added into crossing triangle as OB_SEG.

On the other hand, if DM_SEG looks from the every other part of SILE otherwise to be blocked, then DM_SEG be not actual look from the limit block border (OB) and the treatment scheme row directly marches to step 2370 to determine whether remain any untreated wedge " wedge _ tabulation ".If determine the untreated wedge of residue in the wedge tabulation in decision steps 2370, then treatment scheme is back to step 2350.

On the other hand, if determine not remain untreated wedge in step 2370 in the wedge tabulation, then treatment scheme marches to step 2380 to determine whether there is any untreated triangle in current " traversal " (being connected to current leg-of-mutton set by non-silhouette edge).If determine to have the not triangle of traversal in decision steps 2380 in traversal (being connected to current leg-of-mutton set by non-silhouette edge), then treatment scheme marches to step 2325, and the triangle that wherein next one is not traveled through is processed.

If in decision steps 2380, determine in traversal not the not triangle of traversal, then treatment scheme marches to step 2383, wherein the triangle that intersects with the OB that generates in current traversal is identified and is used crossing OB segmentation that it is carried out again trigonometric ratio.This step can be limited in the 3D trellis traversal boundary that blocks that ought run into subsequently effectively in step 2315, therefore traversal is restricted to the district that is not blocked, and has further strengthened the output sensitive features of the method.

Treatment scheme marches to step 2385 to continue the 3D trellis traversal in the side that is not blocked according to the OB segmentation that travels through birthdate light.Because the wedge from a 2-stream shape can produce the OB point that is positioned on difference (unconnected) the 2-stream shape, the upper traversal that continues of stream shape (part of PAU) of the new connection that this " shelter fusion " that is illustrated in two stream shapes of realizing by the wedge line of OB segmentation place forms.Treatment scheme stops at step 2385 place.

As previously mentioned, in a kind of embodiment of this method, the mesh object that the whole set of triangular mesh object is closed comprises.The mesh object of this sealing similar be generally used for around " day sylphon " for the mesh object of the visual application of interactivity.Figure 23 from silhouette edge to the back-projection approaches of looking the unit, be considered as triangle stream shape will look unit self with mode like the sky sylphon object class.To look the unit and keep current method responsive for output as triangle stream shape, this is because only process the grid that is not blocked (in the SILE-VC axle) that runs into, and each grid in the axle is processed.Certainly, this method naturally identify look unit self determine the district that is not blocked of VSVV and VSVSC according to it.

The alternate embodiments that back projection is processed only adopts SV-ME back projection (looking from the grid silhouette edge) wedge.In the method, the SV-ME wedge with back projection extends to crossing; The plane of the SV-ME wedge of back projection is extended to crossing, simplified the result about looking the back projection observability figure of unit.

In addition, intersect to come conservatively as can optionally eliminating single order SE-MV wedge to intersecting by the plane of extending adjacent SV-ME wedge, can extending to by the plane with adjacent high-order SV-ME wedge, selectivity eliminates the high-order wedge.

Figure 24 A and Figure 24 B show the processing whether " DM_ section " otherwise is blocked from the silhouette edge source of looking of determining of using the structure of the observability figure back projection of looking from silhouette edge that adopts the 3D trellis traversal.

Processing shown in Figure 24 A and Figure 24 B is implemented in the inquiry of calling in the step 2363 of Figure 23.That is, determine otherwise to be blocked whether looking from the silhouette edge source of looking about the DM_SEG that from the back projection structure of single order silhouette edge, generates of the observability figure that looks the unit.

The method shown in Figure 21 A and Figure 21 B is followed in the processing of Figure 24 A and Figure 24 B, determine the DM_SEG that from the structure of looking the unit and looking, generates from the orthogonal projection of looking the observability figure that looks the unit whether from look the source, unit look except in the situation that Figure 24 A and Figure 24 B source be silhouette edge rather than look the unit otherwise be blocked.

Turn to Figure 24, in some embodiments, this test is wherein adopted DM_SEG and the 3D axle between the silhouette edge (DM_SEG-SILE axle) of testing with decision steps 2410 beginnings.

If in decision steps 2410, determine in the DM_SEG-SILE axle, to have the not triangle of traversal, then treatment scheme marches to step 2415, wherein processes to be ended and major part that orthogonal projection 3D trellis traversal is processed jumps to the not triangle of traversal nearer in the DM_SEG-SILE axle.In fact all geometric figures that block potentially of the state that blocks the border that the conduct that this redirect guarantees to affect DM_SEG is looked from SILE are determining that whether border (OB) is identified DM_SEG before and processed from blocking of looking of SILE.

If determine not exist the not triangle of traversal in the DM_SEG-SILE axle in decision steps 2410, then treatment scheme marches to step 2420, and wherein other wedges in DM_SEG and the DM_SEG-SILE axle intersect.This cross-shaped portion can be subdivided into original DM_SEG a plurality of DM_SUBSEG, and each has unified observability of looking from SILE.

Treatment scheme marches to step 2425 to determine current DM_SEG(or DM_SUBSEG) whether directly link to each other with DM_SEG or the DM_SUBSEG of the observability state (otherwise be blocked or otherwise be not blocked) of having determined to look from SILE.

Directly link to each other with DM_SEG or the DM_SUBSEG of the occlusion state of having determined to look from SILE if determine current DM_SEG or DM_SUBSEG in decision steps 2425, then treatment scheme marches to step 2430.

In step 2430, current DM_SEG or DM_SUBSEG are arranged to the state identical with the state of the DM_SEG that has known observability state of looking from SILE and directly link to each other with current DM_SEG or DM_SUBSEG or DM_SUBSEG, and this state is back to call function.

If determining current DM_SEG or DM_SUBSEG in decision steps 2425 does not directly link to each other with DM_SEG or DM_SUBSEG with known observability state of looking from SILE, then treatment scheme marches to step 2435, wherein select the point on the DM_SUBSEG, and treatment scheme marches to step 2440 and is located at the point of selection in the step 2435 and the axle between the SILE with formation, and treatment scheme marches to step 2445, and wherein all wedges in the DM_SEG-SILE axle are identified and are placed in the tabulation " wedge tabulation " to be used for processing subsequently.These wedges are to process in the main 3D trellis traversal that the step 2345 of Figure 23 and 2360(generate orthogonal projection single order wedge) in the wedge of generation.

Treatment scheme marches to step 2448, and wherein each wedge in " wedge tabulation " is processed by subsequent step.

Treatment scheme marches to step 2450, wherein for each wedge in " wedge tabulation ", determines wedge whether single order wedge or high-order wedge.

The main 3D trellis traversal of Figure 23 is processed scanning and the rotary configured single order wedge that generates with step 2345 and 2360 place's wedges.Can process with the orthogonal projection that in the subsequent section of this instructions, discusses in detail in conjunction with Figure 22, Figure 23 and Figure 24 and construct alternatively the high-order wedge.

If determine that in decision steps 2450 wedge is the high-order wedge, then treatment scheme marches to step 2455.

On the other hand, if determine that in decision steps 2450 wedge is not the high-order wedge, then treatment scheme directly marches to step 2458.

In step 2455, by making the high-order wedge with every other wedge and being arranged in wedge-SILE(" wedge-SILE " WEDGE-SILE) UBP of axle and point-SILE axle intersects to construct actual and the corresponding umbra boundary polygon (UBP) of looking from SILE of high-order wedge.For the high-order wedge of identification in step 2450, the part of being positioned at a little of UBP of correspondence-SILE axle is constructed in this processings effectively.Can be by making corresponding wedge and every other wedge/intersect according to corresponding wedge with UBP in the axle of looking between the unit and construct UBP being formed on this wedge.In some embodiments, only construct the subarea of UBP, namely be arranged in a little-part of SILE axle is to answer the polyhedral some pardon test of revising, and it is the later step 2460 of processing.After step 2455, treatment scheme marches to decision steps 2458 to determine that whether the wedge tabulation is as empty.If " wedge tabulation " is not empty, then treatment scheme marches to step 2448, and wherein the next wedge in " wedge tabulation " stands subsequent treatment.

If determine that in decision steps 2450 wedge is the single order wedge, then treatment scheme directly marches to above-mentioned decision steps 2458.

If it is empty determining " wedge tabulation " in decision steps 2458, then treatment scheme proceeds to step 2460 and test to determine that whether it is from otherwise being blocked that SILE looks with the polyhedral some pardon that the DM_SUBSEG point of selecting is made amendment in step 2435.

Figure 25 illustrates this modification the test of polyhedral some pardon embodiment and in conjunction with this figure it is carried out detailed discussion.Briefly, polyhedral some pardon test of this modification determines whether test point is positioned at the PAU inside of looking from SILE.This test is a bit similar aspect the conventional second point (being the point that is positioned on the SILE in this case) of polyhedral some pardon test outside selecting the known PAU of being positioned at.Structure line segment and this line and polyhedral polygon intersect between test point and exterior point.The state of the sub-segmentation of this line (be positioned at polyhedron or outer) can be based on the segmentation of grid polygon with comprise that the cross-shaped portion of the UBP of PAU changes.Needn't intersect to form corresponding UBP(with other wedges because they can not be crossing with other wedges of the side that is not blocked that is positioned at them) the situation of single order wedge under, this test is simplified a little.In addition, in some embodiments because only be positioned at test point-look unit shaft those PAU faces can with the test line segment intersection, do not finish test so do not construct whole PAU.

Treatment scheme marches to from step 2460 and processes 24-1, and it is from step 2465(Figure 24 B) beginning.

In decision steps 2465, the result that the polyhedral some pardon of revising tested checks (step 2460).

If determine that in decision steps 2465 this test point is blocked, then treatment scheme marches to step 2475, and wherein, the result that the DM_SEG that tests is blocked is returned in overall processing.This result is back in the step 2363 among Figure 23.

If determine in the decision steps 2465 that test point is not blocked, then treatment scheme marches to step 2470.

In step 2470, overall process is returned the result that the DM_SEG that tests is not blocked.This result is back to step 2363 among Figure 23.Treatment scheme stops at 2470 and 2475 places.

Figure 24 C: use from the silhouette edge back projection observability figure that looks and construct the conservative visible support that comprises the VSVV corresponding with adjacent silhouette edge and look unit profile contour (VSVSC).

For the sub-segmented construction of specific silhouette edge after the back projection observability figure that looks from silhouette edge, determine VSVV by Figure 22, Figure 23 and processing example shown in Figure 24.VSVV is from the silhouette edge point of looking visible, and this point is to be rotated to the summit on the plane of looking the unit from silhouette edge from the support about the observability figure (and limit or face of looking the unit) that looks the unit that SILE looks.

Process if each single order silhouette edge or its segmentation stand back projection, then it can produce corresponding be incident on identical silhouette edge or sub-segmentation adjusted or high-order SV-ME wedge.High-order SV-ME is supported by VSVV.

Except producing each silhouette edge or the applied VSVV of sub-segmentation, the back projection of looking from silhouette edge also produces as the visible support about the part of the observability figure that looks the unit and looks unit profile contour (VSVSC).

In order to construct the SE-MV that connects adjacent high-order SV-ME, the VSVSC that use produces during the back projection of looking from the limit of the silhouette edge of correspondence.

As previously mentioned, the SV-ME wedge represents to comprise the polygon of the silhouette edge that supports along the limit visibility of blocking on this limit.By contrast, because the light/observability in any source of looking from the district can not be blocked in the summit in fact, so the SE-MV wedge does not reflect the identical aspect of blocking at the place, profile summit that supports.Point can not block any area source.Alternatively, at the inside corner place of profile contour, the observability of looking from the district is comprised/restriction of viewpoint by the light on the face of unit, source/look.The SE-MV wedge that is incident on the correspondence on the profile summit has reflected the restriction that this comprises.

In the single order of this observability method was implemented, generally (with the compound-contoured summit that is regarded as virtual inside corner) located to construct the SE-MV wedge only on inside corner profile summit.Even can constructing the SV-ME wedge by one or more SE-MV limit, high-order SV-ME wedge strictly do not connect at the inside corner place.For example, this situation appears at that single single order silhouette edge is segmented adaptively and determines each high-order SV-ME wedge for every sub-segmentation.

The SE-MV wedge that connects adjacent S V-ME is that the VSVSC from adjacent high-order wedge derives and depends on specifically how VSVSC intersects each other.In the single order situation, silhouette edge is shared the SVSC that shares, and silhouette edge at first segments by the cross-shaped portion with the plane of looking elemental area.

Therefore, the structure of VSVSC and with the set of the SV-MV that determines to connect at the some place that connects two corresponding silhouette edges/sub-segmentation two adjacent high-order wedges from the relation of the VSVSC of adjacent silhouette edge.

In the single order orthogonal projection, the actual support that is used for structure " scanning " single order SE-MV wedge is looked unit profile contour (SVSC) often by 1 simple protruding polyline that forms to maximum about 6 limits looking the unit.

By contrast, during the back projection of looking from silhouette edge is processed, in be connected on the visible support of looking on the unit and look unit profile contour (VSVSC) and can become at random complicated.Depend on the source silhouette edge and look the polygonal layout of grid in the axle between the unit, VSVSC can become non-protruding polyline or or even have a plurality of unconnected polyline of a large amount of segmentations.

Because the purpose of high-order back-projection approaches provide to the simple conservative high-order " adjustings " of SV-ME wedge be connected the SE-MV wedge that connects above-mentioned SV-ME wedge, so a large amount of technology of this method employing minimizes the complexity of VSVSC.These technology can be carried out refinement to improve single order result's precision to the high-order observability in the situation of not introducing too many additional events face.

In embodiment, adopt two strategies to guarantee in back projection is processed, to avoid VSVSC:1) to the identification of following situation: for each VSVSC of the silhouette edge that connects be simple and they simply relevant and 2) the protruding simplification of complicated VSVSC.The first method at first is discussed.

With two VSVSC corresponding to the silhouette edge (or sub-segmentation of the segmentation of same edge) that connects often be simply and they be relevant simply.Two sub-segmentations (A and B) of the identical silhouette edge that consideration is segmented for high-order refinement purpose.The SVV of these two segmentations is identical (definition of supporting plane).The VSVV of segmentation A probably is positioned on the limit of looking the unit.(need not to be identical limit) if two VSVV are positioned on the limit of looking the unit, being used for generating of the VSVSC that then shares, intersect connects A, and be connected the relevant assembly of SE-MV wedge with B be the limit that original SVSC(looks the unit) connect the part of VSVV A and VSVV B.Under this general case, structure connects the SE-MV wedge in the modification of scanning processing, wherein scans between two VSVV.(step 2480 of Figure 24 C and 2483).

In the situation about formerly describing, if it is on the same edge of unit that the VSVV of A and B is positioned at, and these VSVV are produced by the cross-shaped portion that appears at the back projection visibility events face on identical between two parties (back projection) silhouette edge, then segmentation A and segmentation B are positioned in the district of original orthogonal projection silhouette edge, and for above-mentioned original orthogonal projection silhouette edge is by silhouette edge, back projection silhouette edge and look the quadric surface that element sides produces between two parties from looking the umbra event face of looking the unit accurately.The self-adapting subdividing in this district of orthogonal projection silhouette edge produces one group of SV-ME wedge and their conservatively approximate this quadric connection SE-MV wedge, and converges on quadric surface under the limiting case of segmentation.

If VSVV is positioned at (but being produced by the identical silhouette edge of back projection between two parties) on the different edge of looking the unit, then result's SE-MV wedge (together with adjacent SV-ME wedge) is n quadric conservative expression, and wherein n is the quantity of looking element sides between two VSVV.These quadric surfaces are generally connected by the generator limit of being shared.By two contour segmentations of further segmentation and back projection, quadric set of connection can be similar to initial sub-segmentation adaptively.Be connected the described triple tetrahedron wedges of people's (observability bone) or axle test can be used for alternatively seeking accurate connection border, generator limit with Durand by people such as Drettakis, but this is not that this method is required.

The situation of segmentation A and segmentation B is corresponding to source-EEE visibility events quadric surface, by two limits of polygonal mesh with look the event face that a limit of unit forms, by people such as Drettakis described in the description of their completely uncontinuity grid.According to appointment, the situation that this easily identification of back-projection approaches of looking from silhouette edge and high-order, S-EEE event face are corresponding and be similar to these faces with the polygonal approximation of guarding.

Other adjacent silhouette edge/segmentations can cause having and not be positioned on the limit of looking the unit and belong to the VSVSC of correspondence of VSVV of correspondence of the inside of the face of looking the unit.If two VSVSC are protruding and are positioned on the identical faces of looking the unit, and one be comprised in another the inside, then forms the line segment that connects two VSVV.This line segment forms with the profile summit of sharing and scans triangle, and it produces the conservative SE-MV that connects two adjacent adjusting SV-ME wedges.(step 2485 of Figure 24 C, 2487 and 2489).

Although some back projections of looking from silhouette edge produce relatively simple VSVSC, other the back projection of looking from silhouette edge can produce complicated and have not and look on the limit of unit but at the VSVSC of the inner VSVV of one of face of looking the unit.According to each VSVSC of the back projection of the silhouette edge that connects can right and wrong protruding and can have unconnected assembly.In addition, VSVSC can be even not crossing.This so that the operation of scanning at place, the profile summit of being shared be not defined.

In this case, by SV-ME(or the SE-MV of the sub-segmentation support of adjacent silhouette edge) can connect by a series of SE-MV of any complexity, it has reflected the complexity of the VSVC that connects two VSVV.In fact, two VSVV link to each other at the VSVC that can have a more than coupling assembling.

In order to simplify the set of the SE-MV wedge that connects the SV-ME wedge on the adjacent sectional, the VSVSC corresponding with each segmentation carried out conservative simplification.(step 2485 of Figure 24 C and 2491).

This simplification relates to VSVV or visible support of VSVE(sought for every sub-segmentation and looks element sides), then extend accordingly the plane of SV-ME or SE-MV to be divided into two looking the unit by the plane parallel with the wedge that is formed by VSVV or VSVE.

Look this bisection of unit for each adjacent silhouette edge segmentation.Each result's VSVSC has the simple relation with the SVSC that looks the unit: new looks unit profile contour by the original subset of looking unit profile contour of plane cutting.Because original SV-ME wedge is looked cellular construction by the support parallel with each limit and supported, be transformed into the SE-ME wedge so be incident on the original SV-ME wedge of this cutting on the silhouette edge.After the bisection of looking the unit, VSVSC has simple relation usually: one is that another subset is (unless new VSVE intersects looking the unit, this when they corresponding to when the segmentation silhouette edge the time can not occur, this can be parallel in the sort of situation because of them).

Because the SE-ME that adjacent sectional supports looks point-supported on the unit profile contour (SVSC) by support, so SE-ME can be by the conservative profile contour of new VSVSC(by two correspondences of connection of SVSC) the SE-MV that generates of the segmentation of corresponding SE-ME connect.

But, pass the point that the sub-segmentation of profile of two connections shared these SE-MV each can be connected with one of SE-ME of limit sound and can not intersect on the limit with another SE-ME but accurately on the ground.

For seeking two SE-ME, the set that is formed on two all SE-MV between the VSVSC forms according to the triangle that scans of correspondence, and the SE-MV wedge intersects each other and with SE-ME to seek the continuous polygon facet that is connected.Note, these SE-MV wedges can intersect with the SE-ME wedge, are not only Bian-Bian and are to intersect in the inside of wedge.Similarly, SE-MV scans wedge and can also intersect each other in the inside of wedge.

Can adopt the additive method of conservatively simplifying the complicated VSVSC that is generated by back projection.In one approach, non-protruding VSVSC can be processed by any method for the protruding simplification of polyline.The unconnected assembly corresponding with back projection single silhouette edge VSVSC can be by conservatively connecting at the convex closure of looking the unit and form them.Similarly, the unconnected assembly corresponding with back projection adjacent silhouette edge VSVSC can conservatively connect at the convex closure of looking on the unit by forming them.The protruding VSVSC that simplifies produced the simplification that connects the SV-ME that regulates SE-MV.

Connection can reduce the complexity of result's the observability structure of looking from the district (DM, VM or PAU) significantly for the protruding simplification of the shared VSVSC of two VSVSV of adjacent sectional.

In some cases, as seen VSVV for the silhouette edge of two connections looks from the limit (corresponding SV-ME is accurate), and the support part of scanning support triangle shape corresponding with the SE-MV of connection of still looking element sides may be looked from the profile summit of being shared and be blocked.In this case, can be with utilizing the observability figure that looks from point about looking the unit that the profile summit of sharing is processed to generate as the back projection in source to construct VSVSC.The back projection of looking from point adopt Figure 22 and Figure 23 method simplification embodiment.In the back projection situation of looking from point, all wedges are the UBP that looks from point and use the definition of the silhouette edge of looking from point to generate, in the wedge structure, do not use and scan operation and the known OB_SEG of being of each DM_SEG, so do not use the test of Figure 24 A and Figure 24 B.

VSVSC is limited to and looks elemental area.Block the border and in fact can cross over one or more that look the unit, but the continuous polyline that is regarded as shining upon.

In a kind of technology, implement in order to simplify back projection, at first can look the elemental area Plane intersects and come the grid silhouette edge as rear-projection eikonogen is segmented to form segmentation by making.For every sub-segmentation of such segmentation, corresponding VSVSC usually is restricted to the identity set of looking elemental area.In addition, the segmentation of any segmentation can be driven by previously discussed error metrics, and this error metrics is estimated with the deviation of corresponding high-order wedge the single order wedge.

Should be noted that specified back projection of looking from silhouette edge is processed the first order modeling that self has adopted observability to propagate herein.Thereby it is conservative but need not to be accurate adjusting that it has produced the assurance of single order orthogonal projection SV-ME wedge and the SE-MV wedge that is connected them.Can process at back projection and use the high-order thinning method in self.This can be similar to the quadric surface about in the observability figure that look unit corresponding with the described non-luminophor of people (the 1994)-EEE time such as Drettakis.Because these non-source EEE events are seldom facilitated the umbra volume of looking the unit from looking significantly, so the method can be significantly so that implement complicated and the unlikely precision that improves fully single order back projection result.Table I X shows the type (using the nomenclature of Table I) of the event face of explaining with following methods: the back-projection approaches of single order method, the back-projection approaches of using the propagation of single order observability in back projection and use high-order observability.Under latter event, determine that by back projection source (grid silhouette edge) is from (looking from segmentation) grid silhouette edge part of looking visible.As previously mentioned, the E-EV face that the first order modeling that uses in some cases observability to propagate generates is corresponding to the E-EV event face of uncontinuity gridding.In other cases, any event face with uncontinuity gridding method (wherein E-EV event face not necessarily the generates continuous umbra event face) structure that uses prior art is not corresponding for the E-EV face that generates by the single order method.

Table I X

In orthogonal projection and back projection, be similar to the visibility events face of Four types with single order observability transmission method

In a word, back projection process can produce non-protruding and can have the VSVSC of a plurality of unconnected assemblies.This can increase the complexity that is incident on the SE-MV wedge on the summit of being shared by the adjacent sub-segmentation of grid silhouette edge fully.If VSVSC has a plurality of unconnected assemblies, then visibility problem no longer be look from the district but in fact " look from a plurality of districts ".Therefore, be used for control result's observability figure or the complexity of PAU with conservatively connecting the method for a plurality of unconnected assemblies of VSVSC and the protruding short-cut method of single VSVSC.

Again, this method is used specific high-order observability effect (profile retraction, high-order back projection etc.) so that the user can be chosen in from look the observability solution of looking the unit.In addition, can come the use of the high-order effect of specific silhouette edge is instructed based on the heuristics of specific high-order effect by considering the maximum possible deviation between single order result and the high-order result.

Summary for 3D and the 2D trellis traversal algorithm of the observability of looking from the district

That the trellis traversal method of the output sensitivity of the lower structure of 3D situation (PVS that for example, looks from looking the unit) and 2D the situation observability of wedge (for example, about) observability figure is very similar very clearly such as the description according to the front.

In addition, the back-projection approaches that is used for refinement single order observability figure also adopts 3D and 2D traversal method, but uses different sources, shelter and target.

Following table has been summed up 3D with 2D trellis traversal algorithm and has been summed up and how to have adopted different source (vision area), shelter and objective cross that they are used in orthogonal projection and the back projection pattern.To in this sense, target is to construct the geometric figure that observability figure blocks boundary element (point in the segmentation in the 3D algorithm and the 2D algorithm) thereon.

Table X

Modification for the 3D trellis traversal algorithm of the observability of looking the unit

Table X I

For the modification about the 2D trellis traversal of the observability of wedge

Table X II shows face and the p-block element p for the correspondence of 3D and 2D trellis traversal method.

Table X II

Face and the p-block element p of the observability of looking from the district for 3D and 2D

Element	3D	2D
			The umbra boundary element of looking from the source feature	Wedge	Wedge line (WL)
The umbra boundary element of looking from the source	Umbra boundary polygon (UBP)	Umbra boundary line (UBL)
			The umbra frontier district of looking from the source	Polyhedron polymerization umbra (PAU)	Polyline polymerization umbra (PLAU)

Figure 25: use the point of single order wedge and high-order wedge to block method of testing

Whether the method for the linearization umbra observability figure that Figure 19 and structure shown in Figure 20 are conservative is used the point determined on the uncontinuity grid regions is the test of (for example, Figure 19 step 1945 place) of looking and be blocked or look and be not blocked from looking the unit from looking the unit.This point blocks test and is the polyhedral some pardon test of revising and the process flow diagram that is regarded as Figure 25 by formulate.

In embodiment, treatment scheme begins from step 2505, and its middle conductor forms connecting test point (TP) and the point of looking on the elemental area.Supposing to be positioned at this point of looking on the elemental area looks and is not blocked from looking the unit.

Treatment scheme marches to step 2510, and this line segment is with crossing with all mesh triangles shapes, single order wedge and high-order wedge UBP potentially.Mesh triangles shape comprises the test point that is not calculated as cross-shaped portion.

Note, although in this test, can use the single order wedge, at first any high-order wedge must be intersected to form corresponding high-order UBP with other wedges.This is needed, because be different from the single order wedge, the side that is not blocked of high-order wedge can intersect with other wedges.This fact means that the observability about wedge of high-order wedge can have the labyrinth that comprises overlapping hole, and above-mentioned overlapping hole is to be caused by the overlapping wedge that be not blocked side and high-order wedge at the high-order wedge intersect.Thereby, at first determine the observability structure about wedge of high-order wedge by constructing corresponding high-order UBP.Then block at point and use high-order UBP rather than high-order wedge in the test.

Treatment scheme marches to step 2512, and wherein the point on looking elemental area begins and march to test point to come the cross-shaped portion that generates step 2510 is tested.

Treatment scheme marches to decision steps 2515, wherein the type of cross-shaped portion is checked to determine that whether it is corresponding to mesh triangles shape cross-shaped portion.

If determine that in decision steps 2515 this cross-shaped portion is corresponding to mesh triangles shape cross-shaped portion, then treatment scheme marches to step 2520, and wherein the expression value of variable (" LS_ state ") of possible occlusion state of new sub-segmentation that comprises the line segment LS of this cross-shaped portion is set to state and " is blocked ".

If determine that in decision steps 2515 this cross-shaped portion does not correspond to mesh triangles shape, then treatment scheme marches to step 2525 to determine that whether this cross-shaped portion is corresponding to single order wedge or high-order UBP.In decision steps 2525, if this cross-shaped portion is corresponding to the single order wedge, then treatment scheme marches to step 2545 and is considered to be along the side that is blocked of this cross-shaped portion Ying Yu when looking elemental area towards the direction of test point TP and the wedge side that still is not blocked when line segment determining.

If determine this cross-shaped portion corresponding to the side that is not blocked of wedge in decision steps 2545, then treatment scheme marches to step 2550, does not wherein change the value of " LS_ state ".

If determine this cross-shaped portion corresponding to the side that is blocked of single order wedge in decision steps 2545, then treatment scheme marches to step 2555.

If the currency of " LS_ state " is " not being blocked " in decision steps 2555, then treatment scheme marches to step 2560.In step 2560, keep the value of " LS_ state " constant.

If in decision steps 2555, the currency of " LS_ state " is " being blocked ", and then treatment scheme marches to step 2565.In step 2565, the value of " LS_ state " is set to " not being blocked ".

If the structure that determine to intersect in decision steps 2525 is high-order UBP, then treatment scheme marches to step 2530 and determining when line segment LS is considered to be along cross-shaped portion when looking elemental area towards the direction of test point TP is and the cross-shaped portion of the side that is blocked of UBP or corresponding with the cross-shaped portion of the side that is not blocked.

If determine this intersection area in the side that is blocked of UBP in decision steps 2530, then treatment scheme marches to step 2540 so that the value of " LS_ state " is arranged to " not being blocked ".

If determine this intersection area in the side that is not blocked of UBP in decision steps 2530, then treatment scheme marches to step 2535 so that the value of " LS_ state " is arranged to " being blocked ".

When handling all cross-shaped portions, then show the observability of looking from looking the unit of institute's test point by the end value scale of " LS_ state ".Treatment scheme stops at step 2520,2535,2540,2560 and 2565 places.

In embodiment, the some occlusion query of Figure 25 is answered 2D occlusion query and the 3D occlusion query of the selection of depending on source, shelter and target.How the various occlusion queries that Table X III shows as running into during the visibility problem of looking from the district for the 3D in orthogonal projection and back projection (high-order) and 2D use an occlusion query.

Table X III

The modification of 3D and 2D point occlusion query

Figure 26: use the 3D trellis traversal to construct the alternate embodiments of the method for polyhedron polymerization umbra (PAU) according to umbra boundary polygon (UBP)

The alternate embodiments that 3D trellis traversal shown in Figure 20 and the relevant figure is processed adopts UBP to substitute wedge.

In the method, step 2045 to 2060 in each wedge and the secondary wedge of structure crossing to form corresponding UBP with other wedge/UBP that intersect potentially.This is based upon on the following condition: each in the wedge that intersects potentially is also crossing with their wedge that intersects potentially.

Because each UBP is from looking the look border of PAU, unit, so polygonal each the crossing segmentation of UBP and grid is to block border (OB_SEG).

Figure 26 illustrates the embodiment that this substitutes.The 3D trellis traversal method of the method for Figure 26 and Figure 20 A is closely similar.In both cases, initiate traversal with the method for Figure 19.

The method of Figure 26 then in conjunction with the described accurate corresponding step of Figure 20 A until step 2661.For the details of these corresponding steps, referring to the discussion of Figure 20 A.

Step 2661 is following steps: with each wedge and the UBP that intersects potentially intersect with determine wedge which partly be UBP.Those wedges that intersect with current wedge and UBP are arranged in by this wedge and look the axle that the unit forms.If current wedge is the single order wedge, then should construct the wedge (by means of the at first processed step 2610 of the wedge of guaranteeing to block potentially) that intersects potentially.Can recursively carry out step 2661 with all UBP of the current UBP of identification impact.In embodiment, except step 2661 in this case be carry out for single order and high-order wedge, step 2661 is identical with the step 2155 of Figure 21.

In the alternative enforcement side of high-order wedge at current wedge, because the wedge/UBP that is incident on the polygon in this axle may intersect with current high-order wedge, so traversal jumps to wedge-the look nearer not traversal polygon in the unit shaft.

Determine that in follow-up decision steps 2663 DM_SEG(of wedge determines in step 2655) whether also be the segmentation according to the UBP of this wedge structure.

If determine the whether segmentation of corresponding UBP of DM_SEG in decision steps 2663, then treatment scheme marches to step 2688, wherein DM_SEG is added into crossing mesh triangles shape as blocking boundary segmentation OB_SEG.This step is identical with the corresponding step 2068 of Figure 20.Subsequent step in the processing of Figure 26 is identical with the corresponding step of Figure 20.

The method of Figure 26 is determined whether OB_SEG of DM_SEG by the UBP that structure is relevant clearly.Therefore, do not need and to determine that whether DM_SEG separates from looking look otherwise the be blocked test of (Figure 21) of unit.

If adopt UBP, determine then whether point is positioned at any PAU(step 1945) test use polyhedral traditional pardon to test.If adopt single order wedge (as in Figure 20), then step 1945 adopts the polyhedral some pardon test (shown in Figure 25) of revising.

Structure CLUDM(is used the non-output sensitive method of Figure 12) and CLUVM(use Tu20 ﹠amp; The output sensitive method of Figure 15) the estimation that assesses the cost.

Following is used for follow-up equation:

Polygonal quantity in the M=model

N=looks the quantity on the limit in the unit

The quantity of the single order silhouette edge in the S=environment

S _Shaft=the quantity that is formed on single single order silhouette edge and looks the single order silhouette edge in the axle between the unit

M _VThe quantity of the visible polygon the in=model

S _VThe quantity of the visible single order silhouette edge the in=environment

S _VShaft=the quantity that is formed on single single order silhouette edge and looks the visible single order silhouette edge in the axle between the unit

V _wThe quantity on the summit of the cross-shaped portion between=all polygons and the single wedge

M _w=polygonal the quantity of grid that intersects with wedge

V _Svw=about the quantity on (look or look from the limit from point) visible profile summit of wedge

Seg _VwThe quantity about the visible segmentation of wedge of the cross-shaped portion between=grid polygon and the wedge

Following equation hypothesis does not have the single order observability of back projection to propagate.This equation has reflected the estimation that assesses the cost of the observability solution that structure is looked from the district, and they need not to be accurate.

Equation A is the approximate expression formula of using the upper bound that assesses the cost of the conservative umbra uncontinuity grid of conventional uncontinuity irregular triangular mesh design method (Figure 12) structure.The non-sensitive method of output of equation A(Figure 12):

Cost=(N*S*S _Shaft) * LogM/* with all wedges and polygon intersect */

+ (N*S*S _Shaft) V _w* M _w2D when/* uses 2D Weiler-Atherton about the observability * of wedge/

+ (M ²* N ²* S ²* S _Shaft ²) Log (N*S*S _Shaft)/* DM district from look the observability * that looks the unit/

First of equation A is the cost that all mesh triangles shapes and all wedges are intersected.This hypothesis is used and is caused the 3d space hierarchical structure that intersects O (nLogn) to organize geometric figure.

Second is the cost of determining the visible segmentation of wedge-triangle intersect section with the 2DWeiler-Atherton method that the uncontinuity gridding of prior art is adopted.This indication is for each wedge that generates, all summits (V) of intersecting the segmentation of mesh triangles shape can generate must with intersect at wedge (M _w) the ray that intersects of other mesh triangles shapes.M _wIt is (generally very little) subset of all mesh triangles shapes (M).Determine M _wCost express in first.As the table shown in, the quantity of the wedge that generates is N*S*S _Shaft

The 3rd is to determine that for all uncontinuity grid regions that create by gridding this Qu Shicong looks the unit and looks and be blocked or be not blocked.For such test, must test a single point in each uncontinuity grid regions and the line looked between the elemental area for all wedges.Because usually, n wedge and any are intersected at this face and are formed n ²Individual district is so the factor I in this item is the quadratic power of the quantity of the wedge that generated by any silhouette edge.Second factor reflected the quantity of the event face of needs tests.

The 3D of the output sensitivity of equation B(Figure 20 and Figure 15 and 2D trellis traversal method)

Cost=M _v* the 3D axle between the visible triangle of LogM/* and other triangles test */

+ N*S _V* S _VShaft* (Seg _Vw* LogM+V _Svw* LogM)/* 2D about the observability traversal * of wedge/

+ (M _V ²* N ²* S _V ²* S _VShaft ²) Log (N*S _V* S _VShaft)/* from the observability of looking the unit and looking block border polyline */

Equation B is the approximate expression formula of constructing the cost of the conservative linearization umbra observability figure of single order with 3D traversal method of the present invention.

First is triangle-the look untreated leg-of-mutton testing cost in the unit shaft for each mesh triangles shape of accessing/traveling through.This also supposes to organize geometric figure with the 3d space hierarchical structure.

The overall cost of 2D trellis traversal of the observability of the triangle segmentation on the wedge is determined in second expression.Because only on the visual profile limit, generate wedge by the 3D traversal, so cost and visual profile S _VQuantity rather than the quantity of all profile S proportional, as in second of equation A.Use the 2D trellis traversal to determine the cost of the observability on each visible wedge and the visual profile summit (V on the wedge _Svw) quantity and the visible mesh triangles shape of (it produces the ray that intersects with mesh triangles shape M) intersect segmentation (Seg _v) quantity of (it has produced for the 2D axle that checks with the cross-shaped portion of mesh triangles shape M) is proportional.As 3D trellis traversal self, the 2D trellis traversal only generates the intersection (2D axle and ray) that is incident on the visible structure (being respectively segmentation and profile summit).

Last is to determine that it is to look and be blocked or look and be not blocked from looking the unit from looking the unit that the uncontinuity grid that generates blocks boundary segmentation.This point can be blocked the test be used in equational the 3rd point block the test compare.The responsive uncontinuity grid method of a kind of non-output.These two points block the test point of p-wire and the wedge that generates.Serious difference is: for 3D trellis traversal method, required test quantity is the function of the quantity on visual profile limit, rather than the function of the total quantity of silhouette edge.Therefore, for 3D trellis traversal method, the quantity of blocking the wedge that test must intersect for each point is generally much smaller than the required quantity (N*S of art methods _V* S _VShaftVsN*S*S _Shaft).

In usually complicated 3D model, all the quantity of silhouette edges is generally much larger than the quantity on visual profile limit, and usually greater than the visible polygonal quantity of grid.Thereby, comparing equation A, equation B represents more to export responsive cost function.This has reflected the performance characteristic favourable, the output sensitivity of 3D/2D uncontinuity irregular triangular mesh design method.

The estimation that assesses the cost (using the output sensitive method of Figure 26) to structure PAU

As previously described, the UBP structure needs according to wedge relate to the additional step that wedge x wedge intersects.But the final output of this processing is possible by from looking unit one or more (guarding) PAU that the positive surface grids polygon that is not blocked that visible UBP connects forms that looks only tangent.Because do not present non-umbra border in PAU, the PAU of the processing of Figure 26 output is generally simple more than the conservative uncontinuity grid of being exported by the processing of Figure 12.Thereby it need to be better simply polyhedral some pardon test from many less " district " and this test of looking the observability test of looking the unit that the PAU method produces many.

Equation C is the approximate expression that uses the upper bound of the conservative PAU of 3D/2D trellis traversal method construct.

These are:

Polygonal quantity in the M=model

N=looks the quantity on the limit in the unit

The quantity of the single order silhouette edge in the S=environment

S _Shaft=the quantity that is formed on single single order silhouette edge and looks the single order silhouette edge in the axle between the unit

M _VThe quantity of the visible polygon the in=model

The quantity of the visible single order silhouette edge in the SV=environment

S _VShaft=the quantity that is formed on single single order silhouette edge and looks the visible single order silhouette edge in the axle between the unit

V _wThe quantity on the summit of the cross-shaped portion between=all polygons and the single wedge

M _w=polygonal the quantity of grid that intersects with wedge

V _Svw=about the quantity on (look or look from the limit from point) visible profile summit of wedge

Seg _VwThe quantity about the visible segmentation of wedge of the cross-shaped portion between=grid polygon and the wedge

Equation C(is used for the responsive trellis traversal method of output of structure PAU):

Cost=M _v* the 3D axle * between the visible triangle of LogM/* and other triangles/

+ N*S _V* S _VShaft* (Seg _Vw* LogM+V _Svw* LogM)/* 2D about the traversal * of wedge/

+ (N*S _V* S _VShaft) Log (N*S _V* S _VShaft)/* about the cross-shaped portion * of UBP/

+ (M _V ²* N ²* S _V ³* S _VShaft ²) * Log (N*S _V* S _VShaftPolyhedral some pardon test of)/* */

First is at the grid polygon of traversal and looks the polygonal testing cost of not traversal grid in the axle between the unit.This is with identical for first among the equation B of the 3D trellis traversal method of conservative uncontinuity cancellated structure.

Second is to determine about the cost of the observability of wedge and also identical with second of equation B.

The 3rd is that structure wedge-wedge intersects the cost with the layout that becomes UBP and solution UBP according to wedge.In some embodiments, because this layout solves for each UBP, so the order of magnitude of this cost is higher than the order of magnitude with respect to the cost of the uncontinuity grid of correspondence, uncontinuity grid for correspondence, direct being blocked and the layout of visible volume in the Calculation of Three Dimensional space, but only calculate layout on the face of visible grid stream shape.

Last is the cost of determining that whether stream shape grid is blocked by existing PAU.Not by any stream shape district of the overall process traversal of 3D trellis traversal/PAU structure from looking the unit and look fully as seen or being positioned at PAU fully.In the ordinary course of things, the quantity in the stream shape district of traversal is not the uncontinuity grid regions quantity of quadratic power of wedge quantity much smaller than tending to.Thereby the overall cost of the necessary polyhedral some pardon test of UBP method tends to the observability test from looking to look the unit much smaller than the uncontinuity grid regions.

Trellis traversal is implemented: axle intersects optimization and deviation shaft

In the trellis traversal method, travel through a grid " element " at every turn.In a kind of embodiment of trellis traversal method shown in Figure 20, the grid elements that travels through is polygon (triangle).Thereby, in this embodiment, in step 2010 at the triangle that travels through and look and form axle between the unit.Following alternate embodiments can be arranged: the bounding box that centers on the polygonal cluster (the potential thing that is blocked) that travels through is used in the axle test.In this embodiment, at this bounding box and look and form axle between the unit.The grid elements that is positioned at this axle can be blocked the potential thing that is blocked, and the grid elements that pair is positioned at this axle according to step 2030 before to the traversal of the potential thing that is blocked travels through.If this bounding box-look in the unit shaft without any element, then use single bounding box-the look potential occlusion issue that unit shaft solves the whole cluster of the potential thing that is blocked.Because axle intersects the part that test is the interior circulation of trellis traversal method, so these test the performance that any optimization can both promote trellis traversal significantly.

If carefully select the potential polygonal cluster of the thing that is blocked, then can further simplify this axle test.The grid polygon that forms the connection of cluster may block certainly.If can occur in cluster from blocking, then the cluster bounding box can not identified from blocking with the simple axle test of looking between the unit, and this is to appear in the bounding box because certainly block.But if the potential thing cluster that is blocked does not comprise the silhouette edge of looking from the district, then certainly blocking in the cluster can not occur and bounding box can be identified all potential shelters.In a kind of embodiment of trellis traversal method, the polygon cluster is defined as the polygonal connection grouping with the silhouette edge of looking from the district.Can further organize these clusters with the bounding volume hierarchy structure that can further accelerate the axle test.

The trellis traversal method can also be organized into level grouping or element with potential shelter polygon with bounding volume hierarchy (BVH) or other hierarchical space sub-structures.This can also accelerate significantly axle and intersect (and wedge-wedge polygon and other crossing tests).

In the embodiment of trellis traversal method shown in Figure 20, if find potential block element with by the potential matter-element element and look the axle that the unit forms and intersect of being blocked, then traversal " redirect " is extremely potential blocks element.When this redirect occured, traversal then marched to adjacent grid polygon with common flood (breadth-first) pattern of filling.

The another kind of technology that lift shaft intersects the performance of test has adopted deviation shaft.The polygon that is formed by the adjacent polygon on the grid-look unit shaft can be very similar, especially when polygon with respect to looking the unit when very little.In this case, can more effectively carry out axle for a polygon and intersect test, then construct for adjacent polygonal deviation shaft.This deviation shaft is to construct for adjacent triangle with simple the processing.In this was processed, polygon was assumed to be protruding, in fact, is designated as in this embodiment triangle.

Two triangles, triangle A and triangle B share common edge.A and the axle of looking between the unit are called as axle A, and B is like this equally.If A is connected their common edge E and is connected with B, then the result is quadrilateral Q.With the rotation that forms relevant SV-ME and SE-MV support polygon with scan processing and construct in due form Q and look axle (being called as axle Q) between the unit.This axle does not comprise the SV-ME support polygon that is incident on the E of limit, and this is because E is positioned at the inside of quadrilateral and this axle.Be formed on limit E and look axle between the unit (reuse for the rotation of structure support polygon and scan method) and form the border for the volume that shares for axle A and axle B.Therefore, this expression axle B deviation shaft of deducting axle A can be constructed by deducting axle Q from axle B.The support polygon that forms axle Q comprises the support polygon that is incident on triangle A and the B.In fact, those support polygons for the support polygon of A of Q incremental shaft of accurately axle Q and B being subtracted A separates.This relation causes constructing the simple of deviation shaft and effectively processes: B deducts A, has wherein constructed the deviation shaft that axle A(or A deduct previous axle).

For new triangle B, by rotating to construct two SV-ME support polygons on the limit of not sharing with triangle A that is incident on triangle B.Then not all SE-MV support polygons on the summit on summit of limit E by what scanned that structure is incident on B.Identification axle A is incident on single SV-ME support polygon on the E.This polygonal normal vector is reversed.This SV-ME support polygon becomes the border that B subtracts the A axle.Other boundary polygons are the SE-MV support polygons that are incident on the E of limit.These polygons are looked the summit, unit to the support corresponding with other SV-ME support polygons of B and are formed by scanning from the summit of limit E.

This processing makes it possible to by the first front construction of SV-ME(on the counter-rotating limit of being shared) support polygon and this support polygon is connected to new leg-of-mutton SE-MV support polygon comes simply and effectively to construct two adjacent leg-of-mutton deviation shafts.This processing needs below having avoided fully: the general purpose constructing stereo geometric method of calculating the difference of two axles.

Be blocked segmentation and the corresponding single order silhouette edge that Figure 27 A shows support polygon supports the segmentation of out of true wedge

Using the single order rotation of describing in the first embodiment and the wedge that scans method construct is accurate umbra visibility events face, wherein corresponding support polygon or scan triangle and do not intersect with the grid polygon.

On the contrary, if the support polygon corresponding with the SV-ME wedge (or with the SE-MV wedge corresponding scan triangle) intersect with polygonal mesh, then the counterpart of wedge may not be accurate umbra visibility events face.This is because if support polygon and grid polygon intersect, then when from the contour structure that supports of correspondence (for the silhouette edge of SV-ME wedge, with the profile summit for the SE-MV wedge) may block when observing to support and look unity element (for the summit of SV-ME wedge, and for the limit of SE-MV wedge).

Turn to now Figure 27 A, Figure 27 A shows two polygonal mesh objects looking the unit and be labeled as " grid E " and " grid D ".

The single order silhouette edge with respect to looking the unit on " grid D " is labeled as SE1, and this single order silhouette edge is subdivided into segmentation SE1O and SE1U.

The single order visibility events face that is incident on the SE1 is labeled as " wedge 4 " and " wedge 4 " is subdivided into two parts " wedge 4-is accurate " and " wedge 4-is approximate ".

The single order SV-ME support triangle shape that is incident on the SE1 is SP4.SP4 is the whole triangle between a SVV2 and the SE1.

SVV2 is SP4(and " wedge 4 ") support look the summit, unit.SP4 intersects at an INTE place and triangular mesh " grid E ".SVV2 is not blocked when observing from the segmentation SE1U that supports the single order silhouette edge.Therefore, the counterpart of wedge " wedge 4-is accurate " is accurate umbra visibility events face.

When from the segmentation SE1O(O that supports the single order silhouette edge with respect to being blocked) when observing, SVV2 be blocked (by " grid E ").Therefore, the counterpart of wedge " wedge 4-approximate " is not accurate umbra visibility events face but conservative umbra visibility events face.When the support polygon of the correspondence of and if only if single order wedge did not intersect with any geometric figure, the single order wedge was only accurately.Similarly, if the counterpart of support polygon does not intersect with any geometric figure then the part of single order wedge is accurate.

To concern in embodiments of the present invention those accurate parts and those approximate parts of determining the single order wedge as effectively testing.In conjunction with Figure 28 this test is described.In some embodiments, for those coarse parts of single order wedge, the high-order refinement of the visibility events face that develops with one of disclosed back-projection approaches in this manual.

The SE-MV wedge that scans triangle and correspondence is similarly analyzed.In this case, according to some embodiments, use the part that is blocked of determining to look element sides from the some observability method of (looking from point) of looking.Scan the leg-of-mutton sub-segmentation that is blocked and have the out of true segmentation that can come with the high-order back-projection approaches of describing subsequently in this manual refinement of SE-MV wedge.

The support edge SE-ME wedge situation parallel with silhouette edge uses the observability method of looking from segmentation to determine those sub-segmentations of looking from the limit, source and being blocked of silhouette edge.In this case, grid polygon and quadrilateral support polygon intersect not necessarily produce any sub-segmentation silhouette edge from looking blocking that element sides looks.

In any case, if the segmentation on support or supported limit is looked from supported or the strong point be blocked, then the counterpart of single order wedge is coarse.The corresponding element (single order silhouette edge or single order profile summit) of in fact looking from the grid silhouette edge depending on the support element (support is looked the summit, unit or scanned leg-of-mutton limit) of unit in this case, is invisible.

Look under summit, unit (SVV) looks situation about being blocked from the segmentation of the single order silhouette edge of correspondence in support, corresponding single order SV-ME wedge is coarse, but is incident on the conservative expression on the visibility events border in this segmentation of single order silhouette edge.The more accurate SV-ME wedge that is incident on such single order silhouette edge is to identify by the segmentation segmentation and to looking summit, unit (VSVV) for the visible support of every sub-segmentation in some embodiments.

VSVV looks in fact looking visible and be the part of this visible part and the support section between the sub-segmentation of single order silhouette edge of looking the unit from the segmentation of single order silhouette edge of unit.The rotation that is fragmented into corresponding VSVV from silhouette edge produces " adjusting " or " high-order " wedge, and it has conservatively explained the part observability of looking from the limit of looking the unit.

In embodiment, use in conjunction with Figure 22, Figure 23, Figure 24 and the single order back-projection approaches that relevant figure describes and seek VSVV for silhouette edge.

Figure 27 B shows the view identical with the view of Figure 27 A, and Figure 27 B shows by segmentation segmentation and antithetical phrase segmentation and carries out the single order back projection constructs in coarse single order silhouette edge segmentation (SE1O) high-order umbra event wedge with the identification VSVV corresponding with every sub-segmentation processing.

The coarse single order silhouette edge segmentation that is labeled as SE1O in Figure 27 A is subdivided into two sub-segmentation SE1O-A and sub-segmentation SE1O-B in Figure 27 B.

Use sub-segmentation SE1O-1 as line source or linear vision area and then construct the ad hoc approach of single order visibility events face, limit SE2(itself or with respect to the single order silhouette edge of SE1O-A) (back projection) SV-ME wedge of formation.This wedge accurately a SVV2 place with look the unit and intersect.Therefore, for sub-segmentation SE1O-A, VSVV or SVV(SVV2).Thereby corresponding wedge (SV-ME1) is positioned at same plane with accurate wedge " wedge 4-is accurate ".

Use sub-segmentation SE1O-B as line source and then construct the ad hoc approach of single order visibility events face, limit SE2(itself or with respect to the single order silhouette edge of SE1O-B) on again form (back projection) SV-ME wedge.This wedge is with to look the unit accurately crossing at a VSVV1 place." putting A " crossing with the line that illustrates between the VSVV1 and SE2.The corresponding wedge SV-ME2 that is supported by VSVV1 is " adjusting " or high-order wedge.

Point A is the cross-shaped portion of sub-segmentation SE1O-A and sub-segmentation SE1O-B.Construct the SE-MV wedge that is incident on " point " with the method shown in Figure 24 C, in the method shown in Figure 24 C, carry out " scanning " and connect the SE-MV wedge to generate connecting support profiles contour (VSVSC) for the VSVV of the adjacent sub-segmentation of correspondence.In the situation that Figure 27 B, two VSVV(are for the SVV2 of SE1O-A and for the VSVV1 of SE1O-B) be positioned on the actual limit and single SE-MV wedge of looking the unit, SE-MV1 constructs by the processing of Figure 24 C.

This single wedge connects SV-ME1 and SV-ME2 forms continuous umbra event face, and it has been similar to accurate (secondary) umbra event face that is incident on the single order silhouette edge segmentation SE1O better.

Figure 27 C shows the view identical with the view of Figure 27 B except the following stated: now be subdivided into 4 sub-segmentations rather than 2 out of true parts of coming the original single order wedge of refinement by the corresponding segments with the single order profile, to produce being similar to more accurately umbra event face (quadric surface) actual in this district.

Using and the described same procedure of Figure 27 B, is to identify by the observability of looking from the source that will every individual sub-segmentation is considered as line source and carries out single order to look identifying from the sub-segmentation part of looking visible of unit for the VSVV of each the sub-segmentation in 4 sub-segmentations.

In the situation of Figure 27 C, 4 sub-segmentations have the shown VSVV of 4 correspondences.For example, the VSVV corresponding with the sub-segmentation of supporting SV-ME3 is VSVV1.This can be by verifying with the straight flange on the left-hand side that is positioned at SV-ME3, and the line of extension and VSVV1 and SE2 intersect.

Construct connection SE-MV wedge with the scanning processing between the VSVV of correspondence.For example, construct wedge SE-MV2 by scanning between SVV2 and the VSVV2, corresponding support (scanning) polygon is shown as these two points and connects fine rule between the point of SV-ME1A and SE-MV2.

By using the high-order segmentation of coarse single order silhouette edge segmentation, obtained being similar to more accurately accurate umbra event face.

In fact, the method that the observability of looking from sub-segmentation of segmentation out of true segmentation and structure single order is constructed the high-order wedge is equivalent to conservatively be similar to single accurate quadric method, and wherein VSVV is positioned at and looks on the element sides and a limit is between two parties arranged.

Different and front quadric surface building method (for example, point-line-plane parameter method) use this method of single order back projection to guarantee the conservatively secondary umbra event face of approximate exact of the face of constructing.

Figure 27 D shows the figure of identical structure of the structure of Figure 27 A with different visual angles (a little from looking the back, unit), it shows the single order silhouette edge with segmentation SE1U and SE1O and looks as seen from looking the unit single order.

Figure 28 A: by checking single order wedge and the accurate maximum possible deviation between the wedge and controlling the method for the back projection processing of looking from the limit by the coarse segmentation of single order wedge on contour identification limit

Figure 28 show identification single order silhouette edge from the look segmentation that is blocked and determine process flow diagram for the overall process of the VSVV of each segmentation of described segmentation by back projection of its corresponding SVV.

In embodiment, treatment scheme is from step 2805 when running into the single order silhouette edge.This runs into and can occur in during the 3D traversal processing shown in Figure 20, especially in step 2020.

Treatment scheme marches to step 2810 to construct support polygon with processing shown in Figure 4.This step meeting is carried out as the part of the structure of initial wedge, for example in the step 2045 of Figure 20.

Treatment scheme marches to decision steps 2812 to determine whether support polygon exceeds predetermined value (VAL) with the angle of separating between the polygon.The separating plane that is incident on the single order silhouette edge is to form by the rotation to looking the unit of using the opposite sense of rotation that adopts in the structure support polygon.

According to some embodiments, use back projection to process, when the VSVV that in back projection, calculates near be incident on that separating plane on the silhouette edge intersects look the summit, unit the time, appearance can realize that the maximum possible of SV-ME wedge regulates.

This maximum deviation depends on it is the size of unit and the distance that the silhouette edge distance is looked the unit.Usually, especially for the little unit of looking, this maximum angle deviation is along with increasing apart from the distance of looking the unit and reduce, and this is because look the unit apart from the silhouette edge as the back projection light source become more " point-like ".

Therefore, if the angle in decision steps 2812 between definite support polygon and the separating plane is less than the value (VAL) of appointment, processing marches to step 2814, wherein the SV-ME wedge is considered as accurately and wedge is not carried out " adjusting " processing reflecting the high-order observability.

On the other hand, greater than the value (VAL) of appointment, process marching to decision steps 2815 to determine whether support polygon intersects with any grid polygon if in decision steps 2812, determine angle between support polygon and the separating plane.If support polygon does not intersect with any grid polygon, then corresponding single order wedge is accurate and treatment scheme marches to step 2820.

In some embodiments, the value of variable V AL is selected by the user.The high value of VAL can be easy to cause using the single order wedge, and it may not too accurately still can generate quickly.The low value of VAL can make processes deflection generation high-order wedge, and it more accurately still generally takes a long time to construct.

The adjusting of wedge is not carried out in step 2820 expression.

Intersect if determine support polygon and any grid polygon in decision steps 2815, then treatment scheme marches to step 2825, wherein the SSV of grid silhouette edge is determined from its segmentation that is blocked of looking.If support polygon is triangle, then this problem is equal to the segmentation of looking and being blocked from SVV on contour identification limit.This is that the 2D visibility problem that solves is implemented in the simplification that can use in some embodiments 2D trellis traversal shown in Figure 15 to process.In this is implemented, because need the observability solution that look from point accurately, so processing, the 2D trellis traversal in step 1520, adopts from the look profile summit of (SVV) of point.

In some embodiments, in particular cases also adopt the method at SE-ME quadrilateral support polygon.In this case, this processing determine to be supported and to be looked unity element (VCE is the limit in this case) from the segmentation of looking visible of supported silhouette edge.In this case, the silhouette edge definition of adopting in the step 1520 of Figure 15 is looked from the limit.In both cases, by rotation to as in the SV-ME support polygon situation look unit spot or as constructing wedge line for the silhouette edge of SE-ME support polygon.In the SE-ME situation, look comprising from silhouette edge of element sides to gather to the segmentation of looking the back projection observability figure on the element sides for supporting the tetragonal result that should determine.

The sub-fragmented storage that is blocked that to identify in step 2825 is in SEG_LIST.

In subsequent step 2835 to 2855, the segmentation of SEG_LIST is processed.

The treatment scheme row marches to step 2835 from step 2825, and wherein the segmentation (or in SE-ME support quadrilateral situation, the segmentation of element sides is looked in support) that is blocked for its SVV of a joint silhouette edge is configured to the line source for the back projection processing.

Treatment scheme marches to step 2840, wherein constructs corresponding to the segmentation of processing looking the observability figure of unit with the line source of step 2835.In some embodiments, the structure of this observability figure uses at Figure 22, Figure 23, Figure 24 and processes with the 3D trellis traversal for the back projection of looking from the limit shown in the relevant figure.

Treatment scheme marches to step 2845 to determine looking unit profile contour (VSVSC) from the segmentation support of looking visible according to VM.

Treatment scheme marches to step 2850 to determine and to look summit, unit (VSVV) as visible support corresponding to the segmentation of determining support summit VSVSC in step 2845.As previously mentioned, the summit VSVSC that produces the minimum anglec of rotation is VSVV.

Treatment scheme marches to step 2855 to regulate the single order wedge that supported by segmentation so that wedge is not by VSVV but is supported by SVV.

Figure 29 shows by checking single order wedge and the accurate maximum possible deviation between the wedge and being that process flow diagram that the back projection processing of looking from the limit is controlled is come on coarse simple, compound inside corner profile summit by identification for its single order SE-MV wedge.

Figure 29 shows the process flow diagram of overall process that the segmentation of element sides is looked in identification, above-mentioned segmentation from the support profiles summit with look the simple or compound inside corner profile summit of scanning triangle (process on simple or compound inside corner profile summit and look the support polygon that forms between the element sides by scanning) between the element sides and look and be blocked.

In embodiment, treatment scheme wherein runs into the inside corner summit of grid profile contour since 2905 places.As previously mentioned, this inside corner can be that the summit of simple profile contour or it can be as the summit by the crossing compound-contoured summit (CSV) of causing of wedge and grid silhouette edge.Foregoing, determine the inside corner state with the relevant orientation of the silhouette edge that forms the summit, turning.

Treatment scheme marches to step 2910 and scans triangle to come with the processing shown in Fig. 5 A and Fig. 5 B as previously mentioned to look to form between the unit profile contour in inside corner summit and extreme value.This step for example can be used as in the step 2045 and 2060 of Figure 20 initial once with the part of the structure of secondary wedge.

Treatment scheme marches to 2912 to determine whether support polygon exceeds predetermined value (VAL) with the angle of separating between the polygon.The separating plane that is incident on the single order silhouette edge forms to looking the unit by using the opposite sense of rotation that adopts in the structure support polygon to rotate.For the SE-MV wedge, this determines by checking adjacent SV-ME wedge.

Therefore, if the angle in decision steps 2912 between definite adjacent supports polygon and the separating plane is less than designated value (VAL), then treatment scheme marches to step 2914, and wherein the SE-MV wedge is regarded as accurately and wedge is not carried out " adjusting " processing reflecting the high-order observability.Be arranged in by the single order silhouette edge of correspondence and support and look the supporting plane that the summit, unit forms by the adjacent SV-ME support polygon that limits.This supporting plane and separating plane (by around identical single order silhouette edge towards looking the unit but be rotated to form along opposite direction) between angle provided corresponding single order wedge and its corresponding high-order or the maximum possible deviation between the wedge of adjusting.

On the other hand, if determine angle between support polygon and the separating plane greater than the value (VAL) of appointment in decision steps 2912, then treatment scheme marches to decision steps 2915 to determine whether support polygon gives any grid polygon mutually.If support polygon does not intersect with any grid polygon, then corresponding single order wedge is accurate and processes and march to step 2920 and wedge is not regulated with expression.In this case, the wedge that is incident on the silhouette edge is not regulated.

Intersect if determine support polygon and any grid polygon in decision steps 2915, then treatment scheme marches to step 2925, wherein is configured to the back projection observability figure VM of the silhouette edge on shared inside corner profile summit.In embodiment, the structure of this observability figure uses the 3D trellis traversal for the back projection of looking from the limit shown in Figure 22, Figure 23, Figure 24 and the relevant figure to process.If be configured (for example, by the processing of Figure 28) for one or two the VM of correspondence in the adjacent silhouette edge, then treatment scheme marches to step 2935, wherein determines VSVSC according to the VM at step 2925 structure.The structure according to VM of each VSVSC can comprise protruding simplification step in some embodiments.

Treatment scheme marches to step 2940, wherein checks according to the relation between the VSVSC of adjacent edge and to connecting to check for the corresponding SVV of each adjacent edge and/or the conservative compound VSVSC of VSVV.This step adopts the processing of Figure 24 C to determine whether carry out that protruding simplification guarantees to connect the SVV of two correspondences and/or unit profile contour is looked in the conservative visible support of VSVV.

Treatment scheme marches to step 2945 to construct the SE-MV wedge that is incident on the adjusting on the inside corner grid profile summit by scanning the SVV corresponding with adjacent silhouette edge and/or VSVV.This scans conservative simplification that connection is shared, the VSVSC(VSVSC that determines in step 2940) on SVV and/or VSVV.

Result's SE-MV wedge reflection is when partly block the geometric high-order effect of looking the unit when inside corner profile summit is observed.Treatment scheme stops at 2945 places.

Figure 30 A, Figure 30 B and Figure 30 C comprise show among the identification observability figure have that high-efficiency dynamic blocks from looking the look process flow diagram of processing of border, district that the method and conservatively simplifying in the district that is blocked is blocked and corresponding grid profile contour of unit.

3D trellis traversal (Figure 20 and relevant figure) be structure conservative from looking the effective ways of the observability figure that looks the unit.The selectivity of linearization back projection technology (Figure 23 and relevant figure) uses so that result's observability figure (and relevant PVS) can increase in linearization visibility events face departs from the district of accurate visibility events face (generally being quadric surface) fully.This adaptive refinement is take the quantity that increases linearization event face (being that employed observability figure blocks border (OB) therefore also) as cost.

Usually because identification and removed the area of a polygon that more is blocked, so the PVS that derives from the observability figure with high precision computation causes less overdraw during play up working time.

But each OB segmentation of Occlusion Map has also produced the additional polygon that is added into observability figure/PVS, and this is because the again trigonometric ratio of the original triangle that intersects with this OB segmentation.(for a kind of again Triangulation Algorithm, referring to M.de Berg, and M.van Dreveld et.al in " Computational Geometry Algorithms and Applications, Springer is c.1997; page45, its full content merges among the application by reference).Should additional geometric figure be easy to simply to slow down by the primitive quantity that increase is committed to the initial geometry stage of graphics pipeline and play up working time.

The overall cost of playing up in order to reduce working time, embodiment comprise the cost that the rasterizing that is produced by the district that is blocked is reduced and because the cost of the increase that the geometric figure that the additional geometric figure that the district that is blocked introduces causes is processed carries out the method for balance.

The method adopts the heuristics that is called effective static barriers (ESO) to eliminate with being blocked of small size and distinguishes the corresponding district that is blocked, and especially works as them and has introduced relatively large additional triangle.

In relevant method, ESO attempts the surface area that maximization is blocked and distinguishes when also being used for instructing the conservative simplification of blocking the border.

In some embodiments, with the processing of Figure 30 A and Figure 30 C as the off-line observability precomputation carry out.This result is the optimized observability figure that looks from the district and relevant PVS.Proposed with relevant figure in conjunction with Figure 31, Figure 34: by the polygonal certain edges thereof contour of grid being labeled as and the method that has the silhouette edge that is associated of district that is blocked that efficiently blocks and effectively store the structure of observability figure.The silhouette edge that in the method, can use preliminary making calculated for the observability figure that looks the unit/PVS according to the adjacent observability figure that looks (comprising) unit/PVS effectively in working time.The method of Figure 30 A to Figure 30 C comprise not only simplify observability figure block the border and also this information is stored as the original mesh object correspondence simplification the step of profile contour.Carry out the actual Incremental Construction of these method preparation working time observability figure that processes to use Figure 36 and relevant figure/PVS data as precomputation.

In some embodiments, be blocked when what run into observability figure that treatment scheme begins from step 3005 during district (OR).

Treatment scheme marches to step 3007 to determine the leg-of-mutton quantity of the original mesh that fully is blocked in this OR.In the embodiment that substitutes, the leg-of-mutton quantity of the original mesh that is blocked to determining section.Also be in step 3007, come the original mesh triangle that this OR partly is blocked and the leg-of-mutton image-spatial area of original mesh that fully is blocked is constructed with the viewpoint of looking the unit that is arranged in structure observability figure.

Treatment scheme marches to step 3010 to determine the quantity by the additional triangle (being positioned at the district that is not blocked that adjoins current OR) of the leg-of-mutton again trigonometric ratio generation of the original mesh of OR boundary.

Treatment scheme marches to step 3013, wherein adopts in step 3007 and 3010 measured value of determining to determine heuristic variable for effective static barriers of being called as of whole OR (ESO).

In some embodiments, ESO be confirmed as with OR in the proportional variable of the image space area (ISA) of triangle/triangle segments of being blocked.In other embodiment, ESO inspire also comprise reflect rasterizing working time cost this coefficient.

According to some embodiments, ESO is confirmed as becoming certain inverse ratio with the additional leg-of-mutton quantity that is not blocked that produces owing to the again trigonometric ratio of OR boundary in grid.In other embodiment, ESO inspire also comprise reflect geometric manipulations working time cost this coefficient and reflection another coefficient of adding leg-of-mutton storage and transmission cost.Although used the increment PVS working time building method of Figure 36 and relevant figure, can eliminate these additional leg-of-mutton storage/transmission costs by carry out again trigonometric ratio in working time.

According to some embodiments, represent that by following formula ESO:ESO=F(is blocked polygonal quantity, the polygonal image space area that is blocked, 1/ in the quantity of blocking the new visible polygon that boundary adds).In other embodiment, can represent ESO with the function of any expectation.

Other embodiments of this method are possible: it comprises that in fact concrete OR's produces " validity " of blocking of playing up performance working time that promoted supplementary variable in determining generating.

Treatment scheme marches to decision steps 3015 to determine that ESO is whether more than or equal to the value " value 1 " of appointment from 3013.If determine ESO value of being not greater than or equal to " value 1 " in decision steps 3015, then treatment scheme marches to step 3017, wherein removes whole OR and store original, the triangle of trigonometric ratio again not of polygonal mesh from VM.To this, in embodiment, if the ESO that is associated with OR is not greater than or equal to " value 1 ", OR effectively blocks the border.For example, along with the increase of the new leg-of-mutton quantity of passing through again trigonometric ratio processing establishment, the ESO value reduces, and this expresses possibility and does not effectively keep causing again the OR of trigonometric ratio processing.Treatment scheme stops at 3017 places.

Following alternate embodiments is possible: the mesh triangles shape that those that only again are blocked original storage area and keep OR.The triangle that partly is blocked is those additional triangles that produce by the again trigonometric ratio that OR border (OB) locates.By only storing these triangles, reduced the number of triangles among the VM and still from VM/PVS, removed the triangle that is blocked fully.

On the other hand, if determine ESO more than or equal to value " value 1 " in decision steps 3015, then treatment scheme marches to step 3019 to determine whether the leg-of-mutton additives amount of constructing owing to the OR border surpasses " value 2 " in adjoining the district that is blocked of current OR.

If determine that in decision steps 3019 the leg-of-mutton additives amount of constructing is not greater than or equal to " value 2 ", then treatment scheme marches to step 3018, wherein the triangle that produces of current OR(and the borderline again trigonometric ratio by current OR) keep not revising.To this, if additional leg-of-mutton quantity is very little, then can more effectively keep OR.

On the other hand, if definite leg-of-mutton additives amount of constructing is more than or equal to " value 2 " in step 3019, then treatment scheme marches to step 3020.

Step 3020 is implemented following methods to 3026: attempt OR and VM reducing the additional leg-of-mutton quantity that causes in the adjacent district that is not blocked with the adjacent triangle that exposes in the district that intersects on the border that blocks OR by conservatively removing.Use the method, now the triangle that OR blocks boundary that before is limited in the adjacent district that is not blocked is thought not to be blocked fully, and replace these limited triangles with original mesh triangles shape.This again trigonometric ratio that blocks boundary by elimination has conservatively increased from correspondence looks the look area that is not blocked and reduced leg-of-mutton quantity of unit.

Treatment scheme marches to step 3020, wherein " border _ triangle " is identified as those triangles that the border is intersected or adjoined that block with the VM blocked area.

Treatment scheme marches to step 3021, wherein " border _ triangle " removed from OR and with (again trigonometric ratio) border triangle of correspondence from adjacent VM expose that the district removes and with original, larger not again the mesh triangles shape of trigonometric ratio replace.These larger triangles comprise some surface areas that only are included in primitively in the district that is blocked, but not again after the trigonometric ratio step 3021, larger original triangle is thought fully to be blocked, even only have above-mentioned leg-of-mutton part may be arranged in the district that is blocked.

Treatment scheme marches to step 3022, wherein OR and adjacent new border of exposing between the district is arranged to the polyline border that is formed by the adjacent triangle of the original mesh triangles shape with identification in step 3021 of OR.This step conservatively again is defined as the border of OR and is positioned at the OR that calculates primitively.Also reduced potentially the complexity on border and reduced the leg-of-mutton quantity among the OR by conservatively again limiting the OR border.

Follow-up decision steps 3023 is similar with decision steps 3015, and whether conservative again the limiting of blocking the border along the leg-of-mutton limit of original mesh of determining in this case as determining in to 3022 in step 3020 has produced the triangle that blocks sufficient amount and/or blocked the leg-of-mutton blocked area with enough image space areas.

(for example surpassed predetermined value if in decision steps 3023, determine the leg-of-mutton quantity and/or the image space area that are blocked, " value 44 "), then treatment scheme marches to step 3024, wherein blocked area and the adjacent district of exposing is remained on their current state.

On the other hand, if determine the ISA(image space surface area of OR intermediate cam shape) do not have above predetermined value, then treatment scheme marches to and processes 5-1, whether its triangle segmentation level from determine OR has surpassed decision steps 3025(Figure 30 B of particular value (for example, " value 4 ")) beginning.

There is not to surpass predetermined value if in decision steps 3025, determine the triangle segmentation level among the OR, then treatment scheme marches to step 3026, wherein the triangle of OR further segmented (for example, coming to create 4 triangles according to 1 triangle with mid point-Bian segmentation).Follow this process of 3025 test so that extend to also that the triangle of the enormous quantity among the adjacent OR that exposes the district is subdivided into less triangle gradually until the leg-of-mutton quantity in the district that is blocked and/or leg-of-mutton ISA surpass " value 4 " (step 3023) or until the exceedance of levels " value 4 " that segments.Treatment scheme marches to from 3026 and processes 30-2, and it is back to step 3020(Figure 30 A with treatment scheme).

If predetermined value that the triangle in decision steps 3025 among definite OR has segmented exceedance of levels, then treatment scheme marches to step 3027 and blocks the border conservatively to simplify OR.Carry out the ESO that these simplify to increase OR.

In step 3027, the in-order list storage that will form the silhouette edge of profile contour (blocking border (OB) segmentation according to what it constructed the border that comprises current OR) is array " SIL_ tabulation ".For any inside corner profile summit (no matter simple or compound), the unit silhouette edge of looking that is associated of VSVSC also is stored in " SIL_ tabulation ", and this is because they also have and the corresponding wedge of facilitating the OR border.

Treatment scheme marches to step 3028 and is stored among the variable SIL with the initial fragment with array.

Treatment scheme march to step 3029 with the next fragmented storage in " SIL_ tabulation " in variable " next one _ SIL ".

Treatment scheme marches to decision steps 3031 to determine whether SIL and " next one _ SIL " form outer (protruding) turning of profile contour.

If in decision steps 3031, determine SIL and " next one _ SIL " form profile contour (with the OR border of correspondence) outside (protruding) turning, then treatment scheme marches to and processes 30-4, and it begins from the step 3040 at the line segment structure SV-ME on the unshared summit that connects SIL and " next one _ SIL " wedge.Use the VM figure building method of previous appointment, wedge and grid polygon are intersected to form DM_SEG, for from looking the observability of looking the unit above-mentioned DM_SEG being tested to determine whether effective OB_SEG of VM of above-mentioned DM_SEG.These new conservative OB_SEG can intersect with mesh triangles shape and other OB_SEG does not intersect with original OR border.

Treatment scheme marches to step 3042, and the ESO in the district between new OB and the original OB is determined and it is stored in the ESO that variables D _ ESO(represents the poor district between the new and old OB) in.This ESO estimates " validity " of the blocked area that the conservative transposition by OB just has been removed.

Treatment scheme marches to step 3050 to determine that whether D_ESO is less than predetermined value " value 3 ".

If determine D_ESO less than predetermined " value 3 " in decision steps 3050, then treatment scheme marches to step 3060.If the D_ESO in poor district is very low, the leg-of-mutton quantity that then in poor district, is blocked very little and/or they have relatively little surface area.In addition, low D_ESO value can be indicated and have the mesh triangles shape that is blocked by trigonometric ratio again with can causing additional geometric mass part in poor district.

In step 3060, because poor district has relatively low D_ESO value, so the conservative border of recomputating in step 3040 of OR is stored as new OR border for the district of this figure.

Treatment scheme marches to step 3070, wherein alternatively, can also store the triangle of the again trigonometric ratio of introducing by new OR border.The mesh triangles shape that partly is blocked conservatively can be thought be not blocked again, can be passed through again the new geometric amount that trigonometric ratio produces thereby reduced.

Treatment scheme marches to step 3080, wherein will for the data of two silhouette edge SIL that are effectively folded into the conservative silhouette edge of wall scroll and " next one _ SIL " from the expression mesh object be labeled from the chain type tabulation of the profile contour looking to look the unit, remove.Discuss in detail such as the relevant figure in conjunction with Figure 31 and embodiment, the outline data that is labeled is stored as the chain type tabulation of the data structure corresponding with simple profile summit and CSV.In step 3080, the individual node (" contour _ node ") of chain type tabulation is made amendment to reflect the folding of limit.

Be not less than predetermined value if in decision steps 3050, determine the D_ESO in poor district, then treatment scheme marches to step 3084, wherein keep original OB, whether this be because or not do not introducing the again trigonometric ratio of OR boundary in the too many additional leg-of-mutton situation, and poor district is easy to effectively to block the part with relatively large surface area of mesh triangles shape or the mesh triangles shape of relatively large quantity.

Treatment scheme marches to step 3090, wherein SIL is arranged to " next one _ SIL " and treatment scheme and marches to processing 30-5, it is back to step 3029 with treatment scheme, wherein effectively increases " SIL_ tabulation " by the next untreated limit of " next one _ SIL " being arranged in " SIL_ tabulation ".

Form the inside corner of the profile contour corresponding with the border of OR if in decision steps 3031, determine SIL and " SIL_ tabulation ", then treatment scheme marches to and processes 30-3, and it is from determining current contour that processing, corresponding with " SIL_ tabulation " the be blocked outer peripheral line in district (OR) or the interior contour in this district.

If determine that in decision steps 3035 current contour is outer peripheral line, then treatment scheme marches to step 3037 with variable " best side " value of being arranged to " inboard ".

If determine that in decision steps 3035 current contour is not outer peripheral line, then current contour is that interior contour and treatment scheme march to step 3039 so that variable " best side " is arranged to " outside ".

Treatment scheme marches to step 3045, and wherein identification is positioned at two silhouette edges of both sides of SIL or " SIL_ next " so that above-mentioned limit only may near-earth be positioned on the contour (in contour array " SIL_ tabulation ") and so that the corresponding wedge Plane intersects on above-mentioned limit " best side " crossing line to be formed on contour; Wherein " best side " is for " inboard " of the contour of outer peripheral line with for " outside " of the contour of interior contour.This has guaranteed the conservative simplification of the contour in the district of inside corner profile summit (simple or compound).This processing can be to specify span once " to fold " a more than limit by variable N.

Treatment scheme marches to step 3055, wherein determines the ESO in the district between new OB and the original OB and is stored in the ESO that variables D _ ESO(represents the poor district between old OB and the new OB) in.This ESO estimates " validity " of the blocked area that removed by conservative transposition to OB.

Treatment scheme marches to decision steps 3065 to determine that whether D_ESO is less than predetermined value " value 3 ".

If whether determine D_ESO less than predetermined " value 3 " in decision steps 3065, then treatment scheme marches to step 3075.If the D_ESO in the poor district is very little, the leg-of-mutton quantity that then is blocked in the poor district very little and/or they have relatively little surface area.In addition, low D_ESO value can be illustrated in the mesh triangles shape that can be blocked by trigonometric ratio again with causing additional geometric mass part in the poor district.

In step 3075, because poor district has relatively little D_ESO value, thus with the conservative border of in step 3045, recomputating of OR as storing for the new OR border in the district of figure and also being stored alternatively by the triangle of the again trigonometric ratio of new OR border introducing.Again, the mesh triangles shape that partly is blocked can be thought not to be blocked, will pass through again the new geometric amount that trigonometric ratio produces thereby reduced.

Treatment scheme marches to step 3089, will be effectively folded for the quilt of N silhouette edge into the data of the conservative width of the silhouette edge that comprises extension removing and removing from SIL_LIST from the chain type tabulation of looking the profile contour of looking the unit from the mark of expression mesh object.In the discussion in conjunction with Figure 31 and relevant figure, the silhouette edge of institute's mark is stored as the data structure chained list corresponding with simple profile summit and CSV.Subsequently in treatment scheme, in step 3080, can make amendment to represent the limit that folds to the node (" contour _ node ") of chain type tabulation.

Be not less than predetermined value if in decision steps 3065, determine the D_ESO in poor district, then treatment scheme marches to step 3091, wherein because locate the part with relatively large surface area that mesh triangles shape or the mesh triangles shape of relatively large quantity in fact often blocked in poor district, and can not introduce too many additional triangle because of the again triangularization of OB boundary, all keep original OB.

Treatment scheme marches to step 3095, wherein SIL is arranged to " next one _ SIL+N ", and N represents to be processed by the conservative simplification of step 3045 span on the limit that replaces or fold.Treatment scheme marches to processes 30-5, and it is back to step 3029 with treatment scheme, and wherein in fact next untreated limit increases " SIL_ tabulation " in " SIL_ tabulation " by " next one _ SIL " is arranged to.

The method of Figure 30 is applicable to simple and compound-contoured contour.For the compound-contoured contour, inside corner profile summit can be CSV.With the method as after-treatment applications to completed observability figure.Thereby, the silhouette edge corresponding with OB_SEG carried out mark.As previously described, between the tectonic epochs of looking the observability figure that looks the unit, finally the single single order silhouette edge on the triangular mesh can be subdivided into a plurality of visible sub-segmentations.In addition, each segmentation of these segmentations can be supported with another silhouette edge and intersect the SV-ME wedge that scans the CSV that wedge is associated with SE-MV to generate.In the method, " SIL_ tabulation " formed by each sub-segmentation of the visible complexity of the simple or compound profile contour of reflection.

Figure 30 D shows the hiding line chart of 3D of looking the unit and being labeled as two polygonal meshs of " grid F " and " grid G ".This figure has the skeleton view that is arranged in generally the observer and looks the mesh object between the unit.

MESH F is the triangular mesh that expression has the box-like object on six limits.Each rectangular surfaces of " grid F " is modeled as two triangles.12 triangles in " grid F " only show 6, thereby other 6 are blocked not shown.A triangle of " grid F " is formed by the summit that is labeled as V1, V2 and V3.

" grid G " has 21 polygonal polygonal meshs, and some in these polygons represent with triangle, and other polygons then represent with rectangle.In view, some polygons during these are polygonal are blocked.

The 3D that Figure 30 E shows the skeleton view identical with the skeleton view of Figure 30 D hides line chart.In Figure 30 E, show with look the unit as the source from looking the part of the observability figure that looks the unit.Wedge on the single order silhouette edge of structure " grid G " intersects the blocked area that is labeled as OR-G with generation with " grid F ".At this above-mentioned wedge is not shown.Blocked area OR-G is blocked the border of blocking that boundary segmentation forms by 7 and is surrounded.OR-G is positioned at the original mesh triangles shape that is formed by fixed point V1, V2 and V3 fully.

In one embodiment, process to construct the structure that blocks boundary segmentation that surrounds OR-F according to the 3D trellis traversal of Figure 20 A and relevant figure.Particularly, in the step 2068 in the exemplary process diagram shown in Figure 20 A these are blocked boundary segmentation and be added into crossing polygon.

In another step of processing such as the 3D trellis traversal described in the step 2083 in the exemplary process diagram of Figure 20 A, will with block boundary segmentation intersect or comprise block boundary segmentation triangle again triangle change into the Qu Hewei that the is blocked district that is blocked.Figure 30 E show use OR-G block boundary segmentation original mesh triangles shape V1-V2-V3 is carried out the result of this exemplary again trigonometric ratio.At this again during the trigonometric ratio, original triangle V1-V2-V3 is repartitioned into 10 new triangles, it is corresponding one by one that wherein 7 with OR-GE 7 are blocked boundary edge, and 3 other triangles are because 3 districts that OR-G further is subdivided into triangle V1-V2-V3 that summit by OR-G limits and generation.

Because blocked area OR-G is positioned at the single original triangle (V1-V2-V3) of MESH F fully, so it does not comprise completely triangle.That is the polygonal mesh that blocks the border (" grid G ") of, introducing the OR-GE on " grid F " not exclusively block or even will " depending on the unit " as the single grid polygon of " the grid F " in source.The quantity of the mesh triangles shape of therefore, being blocked fully by OR-G is 0.

Triangle V1-V2-V3 is partly blocked by " grid G ".The district of this partial occlusion is the district of OR-G.In this example, the surface area of OR-G is very little with respect to the surface area that comprises triangle V1-V2-V3.

In this example, OR-G does not block any grid polygon fully, and blocks only single polygonal relatively little surface area.Because effective static barriers (ESO) of blocked area becomes certain direct ratio with the polygonal quantity that block fully in the district that is blocked with the polygonal surface area that completely or partially is blocked, all increase the ESO of OR-G deficiently by these factors.

The ESO of blocked area with become the variation of certain direct ratio ground by the new polygonal quantity that creates at the corresponding again trigonometric ratio that blocks boundary.In the example of OR-G, the again trigonometric ratio that blocks boundary has produced and has been divided into 10 new leg-of-mutton single original triangles.

Therefore, in this example, both tend to produce the relatively little value for the ESO of OR-G for the positive specific factor of ESO (the polygonal quantity of being blocked fully and the polygonal surface area that is blocked) and the inverse ratio factor (the new polygonal quantity that for example, is generated by the again trigonometric ratio that blocks boundary).

As described in the exemplary process diagram of Figure 30 A to Figure 30 C, in one embodiment, the OR-G with little ESO can be removed (step 3017) fully from observability figure.Alternately, can simplify blocking the border and redefining for the ESO that blocks the border that has simplified of OR-G according to the exemplary process diagram of Figure 30 A to Figure 30 C.In a kind of short-cut method, the border that is equal to the leg-of-mutton connection collection of the original mesh that is blocked fully in the blocked area by blocking the border is come so that block border less (step 3020 is to step 3022).This cause to as any part exposed the polygon that is blocked again limit.In this case, from eliminating the polygon that only fully is blocked from the PVS that looks the observability figure that looks the unit and development.Simplify in the embodiment at another kind, conservatively simplify the actual border (step 3023 is to 3095) of blocking of blocked area.

The factor of blocking is measured in method employing such as a kind of definite ESO in described a kind of embodiment.In one embodiment, the definite of ESO comprises that also reflection is by the factor of heavily dividing the new polygonal quantity that generates of blocking boundary in correspondence.The method is fit to following embodiment: the relative importance that can regulate based on the coefficient of compound each factor weight each factor.

Principle of operation: increment structure working time of effective storage of the increment PVS information of the silhouette edge of usage flag and observability figure/PVS.

The PVS data are especially if the PVS data of deriving from high precision observability figure can have high carrying cost.As previously mentioned, increment PVS is the art methods that reduces PVS data storage cost by only storing difference between the adjacent PVS that looks the unit.In the people's such as Chhugani (2005) increment PVS method, increment PVS is called increment I for each polygon primitive ID(that looks elementary boundary storage) tabulation.Store by two assemblies with run-length encoding: increment G+(is visible primitive recently) and the primitive that recently is blocked of increment G-() the increment I that forms.Even use this compression, still, increment PVS data are very large.For having 1,300 ten thousand triangles and 500,000 propulsion system of looking the unit, need 7GB to store increment PVS object ID.

Embodiment comprises the method for effectively storing increment PVS with implicit aspect, wherein comes to identify for the concrete silhouette edge that obviously exposes or obviously block of looking unit-look unit transition the generation of model is geometric by relatively corresponding observability figure.Identification and these profile contours that dynamically block or dynamically expose of cousin in the pre-service of off-line observability.This mark adopts the definition that utilizes the single order silhouette edge to realize the effective run-length encoding of obvious compression algorithm.At length provided the trip labeling method in conjunction with Figure 31 and relevant figure.

In working time, be configured to look the observability figure of unit/PVS by following processing according to the adjacent observability figure that looks the unit/PVS:

1) carries out the 3D trellis traversal of the modification of the adjacent observability figure that looks the unit.Fast profile contour that blocks significantly or expose significantly of traversal usage flag is promptly constructed the new border of blocking on the observability figure of structure recently.Provide details in conjunction with Figure 32, Figure 33, Figure 34 and relevant figure.

Use these new borders of blocking, the triangle that recently is blocked is not traveled through during the traversal processing in working time.Because only expressed the triangle of traversal in new PVS, in fact all the method make it possible to generate and have removed the leg-of-mutton PVS that recently is blocked.Provided details in conjunction with Figure 35, Figure 36 and relevant figure.By having eliminated the cost that in most of the cases needs of storing clearly increment G-information (the leg-of-mutton large tabulation that for example, recently is blocked) has been reduced incremental data storage and transmission.

2) geometric figure that recently exposes is looked the related ID of unit transition phase and/or actual geometric increment G+ divides into groups to add as comprising with concrete, the geometric figure that wherein recently exposes " attached " to VM with newly expose the border that is associated.Because can calculate the observability figure that looks from the district to look unit-triangle-segmentation precision, all recently the many triangles in the visible triangle can be the leg-of-mutton segmentation of original mesh.Be adapted at the technology that diabolo between traversal/VM tectonic epochs working time carries out again trigonometric ratio according to this method of the VM Incremental Construction of adjacent VM, it has been avoided must be with the triangle fragmented storage in increment G+ grouping.

In some embodiments, this method is used the unit of looking of level ground layout.Relatively large father looks the unit and comprises less son and look the unit.With previously described Incremental Construction method according to large looking the unit and other increment G+ data are configured to the large VM/PVS that looks the unit for adjacent.Because often very large for the large increment G+ data of looking between the unit transition of looking between the unit, all are the above-mentioned data of cluster naturally, thereby have reduced to load the required magnetic disc access times of increment G+.

Come the VM/PVS that looks (comprising) unit according to the father to derive the VM/PVS that looks the unit for son (involved) by the 3D trellis traversal of making amendment in working time.Because it is the subset that the father looks the VM/PVS of unit that son is looked the VM/PVS of unit, so be used for according to father VM constructor VM working time 3D trellis traversal method only the profile contour that dynamically blocks of usage flag construct and block the border, it walks around the geometric figure that is blocked in the transition of son the uncle.This is so that the more accurate VM/PVS that is configured to show working time in the situation of the increment G+ data that do not need to add.

Also help effective stream processing of increment PVS data depending on the hierarchical organization of unit.In some embodiments, only need to transmit with relatively large father and look increment G+ packet corresponding to transition between the unit.Only use increment G-information (profile information in working time according to mark generates) to come the VM/PVS that looks the unit according to the father to construct the VM/PVS that corresponding son is looked the unit.Because the required overall time of usually searching, access and transmission data units reduces along with the increase of the size of grouping, is more effective so only process father's increment G+ information.

Use the profile information of the mark embody and the increment PVS packet that is associated, the VM/PVS(that the VM/PVS that looks the unit according to adjacent (compatriot) constructs controlled-precision effectively divide into groups with increment G+ and working time the 3D trellis traversal walk around the triangle that recently is blocked).Look unit to son for the father and look the unit transition, can derive the whole VMS/PVS that looks the unit for son because use the silhouette edge hint information exchange walk around the leg-of-mutton mark that recently is blocked to cross the again traversal that the father is looked the VM of unit, so do not need increment G+ grouping.

According to some embodiments, runtime processing is carried out as prefetch operation.In interactivity roaming, be used for the position of current view point based on the connectivity of looking unit (its or the navigation elements of model) and predict possible future when viewpoint speed and other factors and look cell position.Use the speculative prefetching of to know the inside story, increment ground structure and maintain the VM/PVS that the father that can arrive the vicinity when forward sight unit (that is the unit of looking that, comprises viewpoint current reality or that predict) looks the unit.Within the time period of appointment, from being the factor that is tied that comprises the intrinsic navigation constraint of observer's movement when the accessibility set of looking the unit in forward sight unit, comprise viewpoint speed and acceleration and the factors such as governing speed and acceleration such as maximum reality or prediction.The partial structurtes that comprise barrier and other conflict restrictions of modeling environment can also affect the adjacent speed of looking the unit of arrived in the vicinity that can access viewpoint, and this is because structure does not generally need increment G+ data are flowed processing.

This method realized being used for via the flow transmission of the interactivity content of this locality or far-end server effectively based on the codec of observability.This codec utilizes the intrinsic dynamic or interim visible continuity of interactivity roaming to minimize the stream of asking and processes required bandwidth.

Be different from the image-based method for stream processing, visibility events geometric figure and texture information flowed process required bandwidth and relatively be independent of monitor resolution.In fact this method is tended to increase with high resolving power and is played up performance working time, and this is because with the relatively little CPU cost at increment ground structure VM working time, keeps point-device PVS of lifting geometric figure and rasterizing cpu performance.In addition, because this codec may be embodied as speculative prefetching; So it is different from the image-based method for stream processing, its performance relatively is independent of the client-server connection delay.

The data structure that the method for mark silhouette edge of showing Figure 31 A, Figure 31 B and Figure 31 C adopts

As described in the chapters and sections in front, embodiment comprises following methods: identification is dynamically blocked or the grid silhouette edge that dynamically exposes and in the aftertreatment of off-line they are carried out mark; The usage flag silhouette edge realizes in working time the VM/PVS of increment subsequently.

As described in conjunction with Figure 30, can also come the profile contour of looking from the district is simplified based on effective static barriers of blocking boundary segmentation corresponding among the VM.As shown in Figure 30, can be stored as and have simplified mark profile contour (subsequently can according to it on structure is simplified working time VM border) having simplified the VM border.

Figure 31 A comprises the data structure of and compound-contoured contour simple for mark.

In some embodiments, the simple profile contour of triangle stream shape grid is the catenation sequence that comprises the limit of polyline.Polyline can or can not form circulation.

Suppose that simple profile contour is not blocked, then only use single order from the definition of the silhouette edge looking the unit and look; Can come the effectively whole simple profile contour of mark by the single edge of mark contour.The beginning limit of the given single marking data structure of this limit (or point to) can be by seeking fillet and determining which bar fillet is the single order profile contour that the single order silhouette edge is identified whole connection simply.Adopt in the method this fact to reduce the cost of storage mark profile contour by identify most of profile contours limit in working time.

Can be the compound-contoured contour from looking the profile contour of looking the unit.If the visibility events face of looking from the district (for example, UBP) intersects from (different) silhouette edge, then produces compound profile contour.This cross-shaped portion is compound-contoured summit or CSV.

Each inside corner summit of contour no matter be simple or compound, can cause more than one SE-MV umbra event face of looking from the district (wedge/UBP) because scanning to process.Thereby, the more event face that is incident on the contour than the limit in the contour or summit can be arranged.

In some embodiments, will be the array of data structures corresponding with the actual event face that is incident on actual silhouette edge and summit for the data structure organization of mark profile contour.Because the OB_SEG that can promptly identify adjacent silhouette edge and can generate the correspondence of UBP(and observability figure in working time in working time), so in fact do not need to store many elements in the array element.

By identify/generate minimizing that all line elements realize that intrinsic compression algorithm produces in working time the contour nodal information of carrying cost Figure 31 A that can discuss with respect to direct use cost working time that generates this information be balanced.This information is used for generating the not data of storage in acceleration working time.Figure 31 A and Figure 31 B show the embodiment for the data structure of mark profile contour.

In some embodiments, come store data structure " Contour " for every contour.This data structure comprises respectively extremely specifically three fields on the summit on the limit of mesh object, grid and limit.In file layout, all references are the integer index for specific element array, may become pointer although these are quoted.Structure " Contour " also comprises field int node_array, and it is the index to the concrete array of data structures of Type C ontour_Node." struct Contour " also comprises the integer field num_nodes that has provided for the length of the node_array of contour.

Data structure " Contour " also comprises integer field VMinfo, and it is the index for the element-specific in the array of data structures of type VM_Info.VM_info(is described in detail it at the subsequent section of this instructions) comprise information that specifiable lattice is provided and be associated with Contour_Node with mesh triangles shape that UBP intersects.Be stored together by this information of precomputation and with the initial profile element of itself and span, if the UBP that is associated is crossing with identical triangular mesh, then can promptly construct in working time all observability figure OB_SEG that are associated with the whole span of the outline elements of being encoded by Contour_Node.(coming this processing of adopting the intrinsic continuity that intersects polyhedron stream shape is described in detail in conjunction with Figure 35 and relevant figure).

Data structure " Contour " also comprises integer field " last_contour ", and it is that expression current " Contour " in " Contour " structural array is connected to the index of specific " Contour " of its tail end.Data structure " Contour " also comprises integer field " next_contour ", and it is to represent in " Contour " structural array that current " Contour " is connected to the index of specific " Contour " of its head end.

Data structure " Contour_Node " storage is about the information of each element of contour.As previously indicated, because can algorithm ground generate many event faces in the event face that is incident on the contour, so do not need to store clearly them.Thereby, compare limit and umbra event face that actual profile contour has, the Contour_Node array of Contour indication has less element usually.

Data structure " Contour_Node " comprises character type field " node_type ", and it is illustrated in the profile contour information that comprises what type in this node.If node is corresponding to the external corners of profile contour, then the value of this field is set to 1.If node is corresponding and the simple inside corner of profile contour, then the value of this field is set to 2.If node is corresponding to compound-contoured summit (CSV), then the value of this field is set to 3.

Data structure " Contour_Node " also comprises character type field span_type, the span type that its expression is corresponding with node.If node represents that then value is set to 1 for the span of the external corners silhouette edge of contour.If node represents to comprise the span of external corners and inside corner, then value is set to 2." being skipped " is with the span of the silhouette edge of simplifying contour (as in conjunction with the step 3080 among Figure 30 and 3089 described) if node is indicated.

Data structure " Contour_Node " also comprises the integer field of the length of the span that represents representative.Usually, this is corresponding to the quantity that is created on the umbra event face that generates on the silhouette edge that runs in this span and the summit, thereby, can be greater than the quantity on the profile summit in the span.

Data structure " Contour_Node " also comprises integer field ninfo, its be for the storage of array of data structures for the index of the element-specific of the additional information of node, depend on the value of field node_type.

Data structure " OC_Info " can be stored by the additional data with the Contour_Node indication that equals 1 node_type.Character field svsc storage is for the reference of the support of looking the unit corresponding to the index of the certain vertex of the umbra event face (UBP) of single order silhouette edge in the span.But this optional information can accelerate generation working time of UBP increase storage size.Because can come to generate and a plurality of UBP corresponding to unit that look a plurality of in working time with the wall scroll contour, this additional data can once be constructed rather than the profile contour information that serves as a mark is stored with database in working time.

Data structure " IC_Info " can be stored the additional data that refers to by having the Contour_Node that equals 2 node_type.Optional character field ic_type is that expression says that the Constructing Policy that should adopt which kind of type (rotation on adjacent S V-ME plane and scan or intersect) comes that generation is incident on the hint of the SE-MV event face of inside corner sound in working time.Although can make this judgement with previously described heuristics in working time, can avoid test working time with pre-stored data.Note, can also by storage for span_type(value 4 expressions of Contour_Node for the rotation on all the inside corner profile summits in the span and scan crossing for the adjacent S V-ME plane on inside corner profile summits all in the span of vs value 5 expressions) different value nextly store this hint data for the whole span on profile summit.

Data structure " IC_Info " can comprise optional character array field SVSC[4], the specific hint of looking the unit that unit profile contour is looked in the support that its expression forms the SE-MV event face of constructing inside corner profile summit thereon.

All information that comprise in the IC_Info data structure generated in working time, thereby did not in fact need to store with grid database.

Data structure CSV_Info can store by having and equal the additional data that 3 Contour_Node refers to.Integer field " mesh " storage for contour in the index of the specific triangles grid that intersects of current UBP.Integer field " edge " storage is for the index of the certain edges thereof of the triangular mesh that intersects.These two fields be used for to limit the CSV at the cross-shaped portion place that is formed on UBP that the currentElement by the profile contour supports and another silhouette edge.Again, because be easy to determine to intersect to form with UBP the silhouette edge of CSV at (along so that the UBP/VM contour that is associated " leaves " direction of farther grid processes in the situation of current contour) in the half situation, so field " grid " and " limit " are optional.Coming to process in other situations of contour in operation along the direction that UBP and corresponding VM OB_SEG is configured to run into new grid

" mesh " and " edge " field can reduce cost working time of the Incremental Construction of new VM contour fully.

Data structure CSV_Info can also be alternatively be stored in even numbers group field point[3 with x, y, the z value of CSV] in.The CSV_Info structure can also comprise Optional Field char ic_type and char SVSC[4], as described in conjunction with data structure IC_Info.

Because all fields of CSV_Info all are optional, so be not that each Contour_Node can link to the CSV_Info data structure, reduced again carrying cost.

The information that the observability figure that data structure VM_Info storage is associated about the initial profile element of Contour data structure blocks boundary segmentation.Can come to calculate in working time with the storage that the UBP that is associated with these outline elements on grid and limit and VM intersect the corresponding OB_SEG of VM.When having determined OB_SEG, then promptly generate the follow-up OB_SEG that the VM corresponding with the profile contour blocks the border polyline in working time.Being used in this and the CSV_Info data structure accelerate the VM data working time structure grid and the storage class of limit reference seemingly.

The below also reproduced shown in Figure 31 A and Figure 31 B and Figure 31 for the data structure of mark profile contour.

Figure 31 D shows the figure of the data structure of the illustrative embodiments that adopts increment G+ data.

In one embodiment, the increment G+ of information grouping can to look unit-look unit transition phase related with each.Example data structure DeltaGplus_Header comprises that expression begins to look the field that unit (viewcell_end) and certain surface (transition_face) are looked in unit (view_start) and end for the transition that is associated.Another data element DeltaGplus_array is in fact comprising or referring to referring to of the geometric DeltaGplus array of data structures of grid.In some embodiments that DeltaGplus_array are stored in the primary memory, can visit DeltaGplus_array by pointer.In other cases, the DeltaGplus_array variable can be the index of array array, for example, and to the disk storage by increment G+ data formation.Another field DeltaGplus_count stores the quantity of DeltaGplus data structure in DeltaGplus_array.Additional packet_size represents the storage and transmission byte-sized of the information of being associated.

DeltaGplus_Header refers to one or more DeltaGplus data structures, and it refers to again for the geometric figure of the polygonal mesh of correspondence, material, texture information.Example data structure for this information is illustrated as data structure triangle, limit, summit.

Figure 32 A and Figure 32 B show and use the process flow diagram that comes limit and the summit of contour identification contour for the data structure of mark profile contour.

The process flow diagram of Figure 32 show given simple or compound, from the data structure mark that uses Figure 31 of looking the profile contour of looking the unit the situation on some limits under promptly identify the method on all limits of this contour.

The method of Figure 32 is utilized following true: the profile contour generally forms polyline at stream shape grid.Because utilize the link information grid representation that is associated to be digraph (using wing limit data structure or similar data structure), so simplified the identification to the limit that is connected to other limits.

Now turn to Figure 32 A and use the data structure of Figure 31, wherein data element is stored in the array and the index by a little arrays visits these arrays (alternate embodiments can refer to these projection and be pointer working time).In some embodiments, treatment scheme is wherein used from the index contour.vertex for the data structure of current C ontour and is identified current summit c_vertex from step 3205.This is the limit quantity contour.mesh of grid.

Treatment scheme marches to step 3207, wherein accesses similarly with actual contour.edge and works as the front.In this external step 3207, will be arranged to 0 for the integer of the index that upgrades Contour_Node type array.

Treatment scheme marches to step 3209 so that index of reference contour.node_array[ni] the current contour node c_node of access.

Treatment scheme marches to decision steps 3211 to determine that whether c_node.type is not as 3.If the type is not 3, then present node represents that marching to step 3213 for the data of simple contour node and treatment scheme is set to 0 with counter segi.

Treatment scheme marches to decision steps 3217 to determine whether c_node.span_type equals 3.If c_node.span_type equals 2, then the segmentation of current contour span can comprise external corners from the profile summit that the profile summit looking the unit and look and inside corner are looked from looking the unit, and treatment scheme marches to step 3220.

In decision steps 3220, determine with the method on the specified simple profile of identification inside corner summit, front whether the summit of being shared by c_edge and next_edge is inside corner profile summit.

If determine that in decision steps 3,220 two silhouette edges form inside corner, then treatment scheme marches to step 3224.

In step 3224, round values p is arranged to equal to be incident on as construct the quantity of the SE-MV wedge of the upper incident in inside corner summit (step 3228) of determining by using scanning of SE-MV wedge.

Treatment scheme marches to step 3232, will represent that wherein the counter segi for the quantity of the visibility events face of all line widths structure increases the quantity of the SE-ME event face on the CSV that is incident on.

Treatment scheme marches to whether decision steps 3229 equals current contour node with the value of determining segi length of span.

Equal length of span if determine the value of segi in decision steps 3229, then width is treated, and span and treatment scheme march to decision steps 3233.

Determine in decision steps 3233 whether conduct equals the quantity of node in the contour for the value of the integer variable ni of the index of the current contour node of contour.

If determine that in decision steps 3233 present node is last node of contour, then treatment scheme marches to step 3237, wherein next contour is processed.Treatment scheme stops at step 3237 place.

On the other hand, if determine that in decision steps 3233 present node is not last node of contour, then treatment scheme marches to step 3245.

In step 3245, advance the node counts device that in step 3209, uses with the access next node.

If the span_type of definite current present node is illustrated in and does not have the inside corner node on the span in decision steps 3217, then treatment scheme marches to step 3221.

Similarly, if determine that in decision steps 3220 current silhouette edge and next silhouette edge do not form inside corner, then treatment scheme marches to step 3221.

In step 3221, form the VM segmentation corresponding with being incident on single SV-ME UBP on the front (method of using the subsequent section at this instructions of Figure 36, Figure 37 and relevant figure to discuss in detail).

Treatment scheme marches to step 3225 variable segi is increased consistent with the single umbra visibility events face on the structural configuration limit 1.

Treatment scheme marches to the decision steps 3229 of having described from step 3225.

If determine that in decision steps 3221 type of node is the type 3 consistent with compound-contoured summit (CSV), then treatment scheme proceeds to and processes 32-1, and it starts from step 3250(Figure 32 B).

Treatment scheme marches to step 3250 to refer to additional data in the CSV_Info node with being stored in array indexing among the cnode.ninfo.This information has provided the number of grid of the farther grid that comprises CSV.

Treatment scheme marches to step 3255 to intersect the limit quantity on the limit of (wherein cross-shaped portion is current C SV) by CSV_Info access and current UBP.

Treatment scheme marches to 3260 to calculate the CSV as the cross-shaped portion of current UBP and C_EDGE.Alternately, can this value of precomputation and it is stored in the floating-point CSV_Info.point[3 of corresponding CSV_Info structure] in the field.

Treatment scheme marches to step 3265, C_Vertex is arranged to the index on next summit afterwards, CSV summit (being positioned on the side that is not blocked), and treatment scheme marches to and process 32-2, and it is back to step 3215 with treatment scheme.

Generally, the method for Figure 32 A makes it possible to only come with the limit of some marks a plurality of limits of contour identification contour.Thereby this tagging scheme uses considerably less storage.The method utilizes the natural continuity of profile contour to be conducive to promptly to generate the VM segmentation in working time according to the silhouette edge of some marks and the hint information that is associated.Discuss this of structure of silhouette edge of the usage flag of VM/PVS in detail in conjunction with Figure 36 and Figure 37 working time.

Figure 39 A, Figure 39 B illustrate respectively the example shelter be connected from connect look unit A and B and observe the time increment district (DR) of blocking that formed by simple shelter.Figure 33 A be connected with Figure 33 B process flow diagram show convection current shape triangular mesh when from looking the method that is blocked when unit B is observed and identifies from the district of looking the connection that is not blocked when unit A observes.This process provide to determine stream shape triangular mesh when exposing when unit A observes and when the solution of problem scheme in the district of the connection of not exposing when looking the unit B observation from looking.

In some embodiments, looking the unit from and look and expose and look the look district of the connection that is blocked, unit from another in abutting connection with (or involved) and be called increment district (DR) stream shape triangular mesh.

Will with stream shape triangular mesh when from look be blocked when unit B is observed and the increment district of district's correspondence of the connection of be not blocked when looking unit A observation (, be positioned at the observability figure that looks unit A) be appointed as DRoAB(and from A to B, block the increment district).

This with stream shape triangular mesh when expose when unit A observes from looking (, be positioned at the observability figure that looks unit A) and when the increment district of district's correspondence of the connection that is blocked when looking the unit B observation identical.Such increment district is designated as DReBA(and exposes the increment district from B to A).

Certain DRoAB=DReBA.

Therefore, although the method for Figure 33 A and Figure 33 B illustrates the determining of DRoAB (blocking determining of increment district to what look unit B from looking unit A), the handled order of looking the unit is applied to determine to expose the increment district with the method by reversing.

Figure 39 A shows the example polygonal mesh O as the shelter of the larger polygonal mesh among the Occlusion Map 39B.

Figure 39 B shows the observability figure that causes at the face that is labeled as 3910 simple mesh.The 3910th, by the rectangle polygonal mesh that two triangles form, above-mentioned triangle is not shown to simplify this figure.

In some embodiments, the shade hexagon figure that is labeled as " unit A is looked in the blocked area " looks unit A as source and the blocked area of looking from looking the unit that generates with single order observability figure building method in the usefulness of the simple shelter shown in Figure 39 A.Note, because the SE-MV wedge only is created on the inside corner of 5 original limit shelter O, so the limit of blocked area is Duoed one than the limit of shelter O.

Similarly, the shadow-free hexagon figure that is labeled as " unit B is looked in the blocked area " looks unit B as the blocked area of looking from looking the unit of source and use and the generation of single order observability figure building method in the use of the simple shelter shown in Figure 39 A.

" unit A is looked in the blocked area " the outer part that is positioned at of " unit B is looked in the blocked area " is labeled as DR _OAB(blocks the increment district from A to B) and DR _EBA(exposes the increment district from B to A).

Be labeled as 3920 district and be positioned at this outside, two blocked areas, and because it is positioned at from looking unit A and look the unit B district 3910 that looks visible, so distinguish 3920 from looking unit A and looking as seen unit B looks and be not the increment district.

Figure 40 shows the DR that is marked as that is positioned at " unit B is looked in the blocked area " outside except " unit A is looked in the blocked area " _OBA(blocks the increment district from B to A) and DR _EAB(exposes the increment district from A to B) part beyond the observability figure identical with the observability figure that looks from the district of the unification shown in Figure 39 B.Herein, be 4010 with the grid mark that connects observability figure border in thereon, its with Figure 39 in 3910 identical.The black part of " unit A is looked in the blocked area " is divided the part shared with " unit B is looked in the blocked area " of expression " unit A is looked in the blocked area ".Be labeled as the outside that 4020 district is positioned at these two blocked areas, and because it is positioned at from looking unit A and look the unit B district 4010 that looks visible, so distinguish 4020 based on looking unit A and looking unit B as seen and be not the increment district.

Thereby in some embodiments, but when moving from vision area B to vision area A, increment groupings can comprise the poor DR of the visible part between vision area B and the vision area A _EAB(namely, the increment district).Therefore, only comprise the increment groupings in increment district by transmission, owing to do not need again to transmit the whole set of looking the eye diagram element of unit for each, so reduced the bandwidth demand of transmission of graphical information.

Figure 41 A shows identification CSV and the use that is configured to the observability method about wedge (Figure 14, Figure 15 and Figure 16) of the wedge line of SV-ME wedge.

Figure 41 A shows the details about the observability method of wedge of Figure 15.Particularly, Figure 41 A shows the use of the step 1520 of identifying CSV and the use that structure is incident on the step 1545 of the wedge line (WL) on the CSV.

Figure 41 A shows and is labeled as 4101 the top-down front elevation of looking the unit.Some polygonal single order silhouette edges of polygonal mesh are labeled as 4104.Be incident on the single order silhouette edge 4104 the SV-ME wedge by 4104 and limit 4015 and 4106 surround.Use wherein and to look the spinning solution that summit, unit (SVV) is rotated from 4104 to the support that is labeled as 4102 and construct this wedge.

In this example, single order silhouette edge 4104 and look summit, unit 4102 both are arranged in the plane of front elevation.The summit, unit 4103 and 4111 of looking of other marks is arranged in below this plane (it is not clearly at this front elevation).

The SV-ME wedge is crossing from different polygonal mesh objects.Should intersect to have formed and surround 4107 the polyline of being labeled as that is labeled as 4112 inner area (being positioned at corresponding polygonal mesh).

Be labeled as 4108 point and be the profile summit of (look from point in the position in this case) of looking from looking unity element.

Being labeled as in this case 4108 point also by chance is from looking the point on the silhouette edge of looking the unit.

Use the rotation step (that is, rotation is looked the summit, unit to what support the single order silhouette edge that intersects with current wedge) of Figure 15 to construct the wedge line (WL) 4109 that is incident on the CSV4115.

Being labeled as 4108 point is CSV.This point and SV-ME wedge (be incident on 4104 and by limit 4105 and 4106 and delimit) are with corresponding from the cross-shaped portion of looking the single order silhouette edge of looking the unit.This single order silhouette edge passes in and out the plane of this figure.This cross-shaped portion appears at a little 4108 places.

Dotted line 4110 is corresponding to the side view of the single order wedge of the single order that is incident on the plane that passes in and out this figure at point 4108 places from the silhouette edge of looking the unit and looking.This single order wedge is to have to be labeled as the SV-ME wedge that the summit, unit is looked in 4103 support.

Delta between WL4109 and the empty limit 4110 needs not to be the part of wedge.If support wedge look summit, unit (summit 4109) and the single order profile wedge that support to intersect look summit, unit (summit 4103) not in the plane identical with the plane of wedge, then be not arranged in current wedge (by limit 4104,4105 with are connected demarcation) the plane one or more SE-MV wedges can with current SV-ME wedge with at dotted line 4110 places from the side to another SV-ME wedge be connected.

These SV-ME wedges use the method for scanning to form during the 3D trellis traversal is processed (step 2060) at the CSV place.

Dotted line 4113 shows the boundary edge of one of SE-MV wedge of two SV-ME wedges of connection.In this example, by scanning to form the SE-MV wedge from what look summit, unit 4103 to 4111.This SE-MV is connected directly to the SV-ME wedge of sharing limit 4110.This is at a 4108(CSV with current wedge) wedge locating to intersect.

Delimited by dotted line 4113 and another line that is positioned on 4109 by the 2nd SE-MV wedge that scans formation between 4111 and 4102.Three S's E-MV wedge is connected to limit 4109 by scanning between the line above online 4109 and the line 4109 self.In Figure 14 A, regard this SE-MV wedge as double line 4109 from the side.

In this example, three SE-MV wedges SV-ME wedge of being used for intersecting at CSV point 4108 places with current wedge is connected current SV-ME wedge.

In this case, current wedge is that SV-ME wedge and visibility problem are the observabilities of looking from point.In this case, each 2D uncontinuity net point about wedge (wedge line and the polygonal cross-shaped portion of grid) with from the VCE(point) look to block frontier point corresponding.Therefore, do not need additional testing in the execution in step 1563.

Figure 41 B shows the figure that processes about the observability of wedge, comprises for connecting the wedge line structure of SE-MV wedge in the situation that the profile summit is located and polygonal mesh intersects at the tip that forms single order profile contour.

Figure 41 B shows the single order grid silhouette edge (be labeled as 4123) identical with the single order grid silhouette edge (being labeled as 4104) of Figure 41 A.SV-ME wedge among two width of cloth figure is accurately supported by identical point (be labeled as 4121 in Figure 42 B, be labeled as 4102 in Figure 42 A).Because the strong point is identical with supported silhouette edge, so result's SV-ME wedge is also identical.

In the situation that Figure 42 B, depending on looking the large of unit than Figure 41 A in the unit.

In the situation that Figure 42 B, from looking profile summit (4127) that unity element looks not with corresponding from looking the single order silhouette edge of looking the unit.Thereby, by intersecting at this place, summit by face and the polygonal mesh that scans the SE-MV wedge one-tenth of processing on this summit of structure.This cross-shaped portion is the tip of single order profile contour.Figure 41 D shows the tip (being labeled as CUSP) of the single order profile contour among the 3D, in conjunction with figure details is discussed.

In the example of Figure 42 B, between WL4129 and line 4131, form the SE-MV wedge.This SE-MV supports current wedge by looking summit, unit 4121() and the scanning between the summit, unit 4122 of looking that is supported on the SV-ME wedge on the single order grid silhouette edge on plane of figure of grid vertex 4128 places turnover of mark form.In other cases, this SE-MV wedge can not be arranged in the plane of current wedge.Clearly, this SE-MV wedge at polygonal mesh in intersecting from looking the profile summit 4127 that unity element looks about wedge.Thereby, the uncontinuity net boundary occurs at polygonal mesh 4126, and a segmentation on this border comprises or the summit 4127 of the internal vertex of current wedge.Therefore, the point 4127 be positioned at polygonal mesh visible part point and be positioned on the single order profile contour initial point both, thereby make it be called the tip of contour.

Dotted line 4130 and the plane that is formed on single order silhouette edge 4128(discrepancy figure) on the SV-ME wedge corresponding.Usually this wedge is connected to current wedge from looking the unit part of looking visible by the uncontinuity net boundary that is intersected to create by SE-MV wedge (4129-4127-4131) and polygonal mesh.In this case, being positioned at uncontinuity net boundary on the face of polygonal mesh connects two other unconnected single order profile contours and forms continuous umbra event face.

The processing about the observability of wedge for the SE-ME wedge that Figure 41 C shows that the 2D trellis traversal of using Figure 15 and relevant figure processes.

The 4165th, depending on the unit.In this case, the limit with end points 4163 and 4167 depending on the unit is to look element sides for the support of single order grid silhouette edge 4172.In this case, accurately parallel and formation SE-ME wedge of element sides is looked with support in limit 4172.This SE-ME wedge is comprised of single order silhouette edge 4172 and line 4174 and 4176.Rear two lines are to extend to form by the diagonal line limit to the SE-ME support polygon of correspondence.

Three mesh objects 4170,4188 and 4186 and this wedge intersect.The single order UBL that is incident on the single order profile summit is shown as dotted line 4178,4190 and 4192.

Grid 4188 intersects at point 4193 places with the border that comprises of wedge.In a kind of embodiment that the 2D trellis traversal is processed, the some place of intersecting at the wedge (for example, sharing another wedge on limit 4147, not shown) with first pre-treatment initiates traversal.In this case, grid 4188 comprises the beginning segmentation that 4193 ambiguity line segment a little is the 2D traversal.

Traversal marches to and is labeled as 4194 next segmentation.The axle that segmentation 4194 and support are looked between the element sides comprises the segmentation of grid 4170, so end to jump to 4170(step 1510 to the traversal of grid 4188 and with traversal).

During the traversal to grid 4170, construct WL4178 and 4192 in correspondence from looking the profile summit (step 1520 and 1545) that element sides looks.

For among the WL each, throw to seek cross-shaped portion (step 1555) with nearest polygonal mesh object by ray.Being marked as these cross-shaped portions of 4182 and 4184 is respectively the potential frontier point that blocks.

Each point in these points is tested to determine that it is from looking unity element (VCE) whether otherwise be blocked (step 1563) of looking as the support on limit in this case.The 2D version (polygonal some pardon test) that this test adopts polyhedral some pardon of the modification of Figure 25 to test.

In the situation that point 4182 and 4184 uses the segmentation of grid polygon and other single orders WL(to comprise 4190) polyhedron by revising/polygonal these points of pardon test proof are to look and otherwise be blocked from looking the unit.Therefore, above-mentioned point is not to block frontier point and do not limit traversal.

The final traversal of on the limit 4194 that is found to have the profile summit, again initiating.In step 1545, at this profile summit (and the silhouette edge of looking from looking the unit) structure WL4190 that looks from looking element sides.In step 1555, the cross-shaped portion of this WL and nearest grid ambiguity line segment is defined as a little 4180.

In step 1563, will put 4180 and be defined as looking element sides from support and look and be not otherwise as seen, from but block frontier point.This point prevents from traveling through the part that is blocked that enters into 2D grid 4186.This part that is blocked is labeled as 4195.

Traversal is finally advanced (step 1583 and 1585) until run into and comprise frontier point 4196 in the side that is not blocked of OB point 4180.

Figure 41 D shows polygonal mesh, looks the part that comprises tip and compound-contoured summit of unit and single order profile contour.This polygonal mesh is labeled as " grid B1 ".In one embodiment, " grid B1 " is stream shape triangular mesh.In Figure 41 D, some limits in the not shown grid polygon limit are with reduced graph.

Three single order silhouette edges and their the SV-ME wedge that is associated (supporting the summit wedge) are carried out mark.The whole semiinfinite scope with these wedges does not illustrate these wedges.Wedge SV-ME-WB2 is incident on the single order profile E2.Wedge SV-ME-WB3 is incident on the single order silhouette edge E2.These two wedges all are to be supported by the summit, unit of looking that is labeled as SVV-B2.

The part of another single order silhouette edge is shown the grid vertex that is labeled as VB and than the line segment between the compound-contoured summit that is labeled as CSV-B.In this view, another part of this single order silhouette edge is blocked.(it is blocked by wedge SE-MV-WB12).Wedge SV-ME-WB1 is the support summit wedge on the whole that is incident on this single order silhouette edge.Wedge SV-ME-WB1 has the corresponding support that is labeled as SVV-B1 and looks the summit, unit.

The grid vertex that is labeled as CUSP-B is the inside corner summit of MESH-B1.The support edge wedge that is labeled as SE-MV-WB12 is incident on CUSP-B upward and has the corresponding support that is labeled as SVE-B12 looks element sides.

The grid polygon limit that is limited by the line segment between summit VB and the summit CUSP-B is not the single order silhouette edge, this be because the grid polygon of sharing this limit not with respect to all " depending on the unit " back sides, summit pair.Therefore, comprise that limit E3 is connected the chain of single order silhouette edge of connection and finishes at the CUSP-B place with E2.

But, be continuous by wedge and by the conservative linearization umbra event face that they and the polygonal cross-shaped portion of grid form at the CUSP-B place.In this case, continuity is to be created by the wedge SE-MV-WB12 that is incident on the CUSP-B.Wedge SE-MV-WB12 and grid polygon intersect the boundary segmentation that blocks that is labeled as OB-B with formation.Wedge SE-MV-WB12 also with the line that is formed by VB and CSV-B on the single order silhouette edge intersect.The cross-shaped portion of wedge SE-MV-WB12 and this single order silhouette edge is compound-contoured summit CSV-B.At the CSV-B place, the segmentation (not shown) that wedge SE-MV-W12 is divided into the single order silhouette edge that intersects the segmentation that is not blocked (segmentation that is limited by VB and CSV-B) and is blocked.

Three line segment E2, OB-B and CSV-B are supporting the continuous single order umbra contour of " grid B1 " formation of corresponding continuous linearization umbra event face (wedge SV-ME-WB2, SE-MV-WB12 and SV-ME-WB1).

It gives prominence to " top " at wedge SV-ME-WB1 in this view with wedge SE-MV-WB12() intersection that is labeled as " limit 1 " is shown into the cross-shaped portion of SV-ME-WB1.Construct the embodiment of the PAU corresponding with these CLUES and determine this cross-shaped portion.Other embodiments can be determined this cross-shaped portion ambiguously such as the embodiment of the observability figure that structure is looked from the district.

Figure 33 A, Figure 33 B, Figure 33 C and Figure 33 C comprise show to expression from a method of looking to identify to the VM region of interest (ROI) the unified observability figure of the relevant transition of looking the unit unit and for to the process flow diagram that profile contour corresponding to border carries out the method for mark that blocks with those ROI of efficiently blocking.

Figure 33 A, Figure 33 B are connected the embodiment of the method in the district that identifies the connection that is called region of interest or ROI of flowing the shape triangular mesh with Figure 33 D, ROI has reflected the variation of the observability during looking unit transition (for example, from VC A to VC B).

This process flow diagram also shows the method for the identification profile contour corresponding with the border of blocking border (OB) or restriction ROI.In addition, this process flow diagram shows these profile contours of mark (using the Contour data structure of Figure 31 and the data structure that is associated) and the additional information that is associated with contour (for example, the VM_Info data of Figure 31 A) of storage.

According to some embodiments, the method for Figure 33 A to Figure 33 D is carried out as off-line or pre-calculation process.This method can be identified increment G+ and the increment G-assembly of increment PVS data, can use said modules during the runtime processing according to the new PVS of existing PVS increment ground structure.Alternately, the method for Figure 33 A to Figure 33 D can identification and the mark profile contour corresponding with the border of ROI.Can come to construct in working time increment G+ and increment G-with this mark profile contour information subsequently.Compare clearly storage/transmission increment G+ and increment G-grouping, the method can need less storage/transmission resource significantly.

Look in the unit if be completely contained in another depending on one of unit, then ROI is corresponding to only from the involved district that looks visible, unit of looking.In some embodiments, come to come constructor to look unit VM(at the VM that look the unit according to the father working time with the mark profile contour corresponding with the ROI of these types and use 3D trellis traversal hint, that simplify), thus clearly storage to increment G-information avoided under specific circumstances.

If look the unit and share face and one and look the unit and be not comprised in another and look (relation that is called " adjacency ") in the unit for two, then ROI is corresponding to increment district (DR).For looking unit transition (being called the AB transition) from VC A to VC B, DR has two types.One type increment district DR _OAB(blocks the increment district from A to B) comprise from looking unit A look visible from looking the unit B mesh triangles shape segmentation of looking sightless.DR _OAB also is DR _EBA(exposes the increment district from B to A).Similarly, DR _OAB=DR _EBA.DR _OAB is corresponding to the increment G-data for the AB transition, and DR _EBA is corresponding to the increment G+ data for the AB transition.Can use the mark profile contour corresponding with the border of these ROI come working time construct increment G+ and or increment G-data (also using 3D trellis traversal hint, that simplify), thereby avoided under specific circumstances clearly storage to increment G+ and increment G-polygon segment information.

Now turn to Figure 33 A to Figure 33 D, in the phase one of processing, specify for the type of specifically looking the ROI that the unit transition will identify.ROI is the given zone that concrete unified observability denotation of graph is looked the unit transition.To comprise from looking unit A and looking look visible mesh triangles shape segmentation and for each the observability figure from the OB_SEG that looks the VM that looks the unit that looks the unit of unit B for the unified observability figure of AB transition.

In some embodiments, treatment scheme is looked the unit below determining and whether is had father and son's (comprising-involved) relation: determine the increment visibility information for the transition of concrete unit from step 3305 in this step.

If determine to have set membership for the unit of looking that will determine the increment visibility information in decision steps 3305, then treatment scheme marches to step 3307.

In step 3307, the ROI that identify is from look the district that looks visible, unit for concrete involved (son) of looking the unit transition.Because the VM that looks the unit for son is father's subset of looking the unit always.Come constructor VM so use in some embodiments the clear and definite increment G-information of calculating for transition.But, look the unit such as fruit and look the unit less than the father significantly, then corresponding increment G-information can be relatively large.Alternately, in the method, can look those districts that look visible, unit from son with only being arranged to for the region of interest (ROI) of such transition.By identifying seed triangle and the border in these districts, can determine by the VM that 3D trellis traversal simplification, hint of unified observability figure being carried out begin with the seed triangle is looked the unit with still less information according to the father that usually son looks the VM/PVS of unit.

If determine to determine that the unit of looking of increment visibility information does not have set membership in decision steps 3305, then treatment scheme marches to step 3309.

It is adjacency (sharing face and one depending on the unit looks the unit coal and be not comprised in another and look in the unit) that the unit is looked in two of step 3309 expressions.Father and son and syntople are two two special arrangement of looking the unit, wherein look the unit from one and appear to another transition of looking the unit and be comprised in two spatial volumes the associating of looking the unit fully.Use these two to look unit step, guaranteed as the ROI that in processing, constructs subsequently reflected fully and conservatively recently as seen and the district of recently exposing (for example, because do not have " gap " in relevant looking between the unit, so missed of short duration visible geometric figure).

It is observability increment district (DR) with expression ROI that treatment scheme marches to step 311.

Treatment scheme marches to decision steps 3313 to enumerate all kinds of DR.Look visible and look invisible from looking unit B from VC A if determine the DR that will identify in decision steps 3313, then treatment scheme marches to step 3315.

In step 3315, the ROI that identify is DR _OAB and DR _EBA.

Look visible and look invisible from looking unit A from VC B if determine the DR that will identify in decision steps 3313, then treatment scheme marches to step 3317.

In step 3317, the ROI that identify is DR _EAB and DR _OBA.

In some embodiments, step 3305 to step 3317 depends on that transition looks the intended use of relation between the unit and increment visibility information and only enumerate the type of looking the ROI that the unit transition will identify for concrete.For in abutting connection with the concrete transition along the AB direction of looking between unit A and the B, usually identify this two class ROI, DR _EAB and DR _OBA.This two class ROI has described fully together for the observability of looking the unit transition and has changed (increment VM/ increment PVS).

Beginning from step 3319, the actual identification in these ROI districts among the unified observability figure of beginning.

In step 3319, determine for the VM(that looks the unit that comprises the associating of looking unit A and looking unit B for example to use previously described any method from look the VM building method of looking the unit).Any superset that can also use this VM as structure subsequently with look for two of transition the unit from looking unit OB_SEG(step 3321 corresponding to observability figure of looking) beginning VM.This fact makes it possible to use the effective level of the visibility problem of looking from the district to decompose and solves determining of increment visibility information (the contour data of clear and definite increment G grouping or corresponding mark).In this hierarchical method, can will look visible triangle segmentation as the VM that is configured to look unit A and looks unified the VM of unit B from comprising any unit of looking of looking unit A and looking unit B.(can also reduce increment PVS storage demand with this fact, this be because for many increment PVS that look the unit transition finally can according to comprise other data of looking single unified observability figure corresponding to unit of looking the unit and generate).

Treatment scheme marches to step 3321, wherein from look unit (A+B) look visible triangle segmentation structure with according to look unit A determines from look OB_SEG corresponding to the observability figure that looks the unit and with according to look unit B definite from looking OB_SEG corresponding to the observability figure that looks the unit.To look visible triangle segmentation together with from looking OB_SEG that unit A looks and being called as for looking unit A and looking unified the observability figure of unit B from the set of looking the OB_SEG that unit B looks from looking unit (A+B).In some embodiments, the structure of these OB_SEG adopts previously described for constructing 3D/2D trellis traversal method (Figure 20 and relevant figure) from looking the VM that looks the unit.

Certainly, if be the father who looks unit B depending on unit A, in step 3319 structure from look the observability figure that looks unit (A+B) comprised from look unit A look visible all mesh triangles shape segmentations and with from looking OB_SEG corresponding to profile contour that unit A looks.In step 3321, add the OB_SEG corresponding with looking unit B.Look cell abutment and/or employed beginning VM is VM(A+B at two) the situation of superset under, generally must two OB_SEG set of structure in step 3321.

Treatment scheme marches to from step 3321 and processes 33-1, and it is from step 3323(Figure 33 B).In step 3323, unified observability figure is traveled through to limit by from looking unit A and looking the OB_SEG that unit B looks and arrange the VM district that forms.In this traversal, select triangle and traversal to march to the border that is formed by OB_SEG.In this step, the 3D trellis traversal is carried out as the breadth first traversal of the unified observability figure that has constructed.Traversal marches to the border of blocking of the limited profile contour of traversal and correspondence.This simple traversal method has guaranteed that all the triangle/segmentations to concrete ROI travel through the border to ROI, even ROI has internal holes or crosses over a plurality of independent triangular mesh.(the single ROI that the structure of the front of VM finally can become have internal holes with the part " fusion " of triangular mesh, above-mentioned ROI corresponding to farther mesh triangles shape, pass the visible district that is not blocked, hole in the nearer triangular mesh that is surrounded by interior profile contour).

Treatment scheme marches to step 3325 to determine whether remain any not triangle of traversal in current traversal/VM district (that is, be connected to the current leg-of-mutton not triangle of traversal, wherein connect not need to pass and block the border).Determine to have the not triangle of traversal in current traversal in step 3325, then treatment scheme marches to step 3327, the next triangle in the traversal.

On the other hand, if in decision steps 3325, determine in current traversal, there is not remaining triangle shape, then because traveled through all triangles of current VM, so treatment scheme marches to step 3328.

The unified VM district that determines identification in traversal step 3323 and 3325 in step 3328 looks from looking unit A, or looks from looking unit B, or looks as seen from both.In some embodiments, test with polyhedral some pardon of the Figure 25 that has simplified this is determined.In the relevant situation of looking the unit transition of father and son, so look and as seen can carry out some simplification to this test because the district uncle of all traversals looks the unit.

Following alternate embodiments is possible: initiate at first a single point in the VM district to be determined (step 3328 and 3331) before the traversal (step 3323) in the VM district.The method make it possible in the situation that not to the VM district entirely travel through identification be not the VM district of ROI.

Treatment scheme march to decision steps 3331 with the district of the traversal of determining VM whether corresponding with before for specifically looking the unit transition in perception district (ROI) that previous step 3305 is established in to 3317.This is by as seen the result of step 3328 (for example, is looked, looks visible or look visible or look invisible from both from both from B from A; The latter only when employed VM be VM(A+B) superset the time just might) compare to determine with the restriction of the ROI that determines in to 3317 in previous step 3305.

If determine that in decision steps 3331 district of the traversal of unified VM is not ROI, then treatment scheme marches to decision steps 3349 to determine at VM whether any not triangle of traversal being arranged.

If determine any not triangle of traversal of residue in unified observability figure in decision steps 3349, then treatment scheme marches to step 3337, wherein select among the unified VM next triangle (belonging to new VM district) with for the treatment of.

If determining in decision steps 3349 does not have the not triangle (no longer including unified VM to be processed district) of traversal of residue in unified VM, then treatment scheme marches to step 3352.Treatment scheme stops at 3352 places.

If determine that in decision steps 3331 district of the traversal of VM is ROI, then treatment scheme marches to step 3332.

Determine that in decision steps 3332 whether current ROI blocks increment district (DR_O) for what look the unit transition.If ROI is DR _O, then treatment scheme marches to step 3334.

In step 3334, determine effective static barriers of ROI with the tolerance of before describing for the district that is blocked.The value that will be called the variable of effectively dynamic observability (EDV) is arranged to the ESO of ROI.

If determine that in decision steps 3332 current ROI is not DR _O, then treatment scheme marches to step 3333.

Determine that in decision steps 3333 whether current ROI blocks increment district (DR_E) for what look the unit transition, if ROI is DR _E, then treatment scheme marches to step 3336.

In step 3336, determine effective static barriers (ESO) in the district that is blocked (being called around district or the SOR of being blocked) around current ROI with the tolerance of before describing for the district that is blocked.The value that is called the variable of effectively dynamic observability is set to the polymerization ESO of SOR.

If determine that in decision steps 3333 ROI exposes the increment district, then treatment scheme marches to step 3335.

Determine that in decision steps 3335 current ROI is whether corresponding to look the unit district correspondence of looking visible from the son of looking the unit transition for specific father and son.If ROI is the subarea, then treatment scheme marches to step 3338.

In step 3338, determine effective static barriers (ESO) in the district that is blocked (being called around district or the SOR of being blocked) around current ROI with the tolerance of before describing for the district that is blocked.The value that will be called the variable of effectively dynamic observability is arranged to the polymerization ESO of SOR.Note, for DR _ECarry out identical processing with the situation of sub-ROI, but in order to illustrate, they are different herein.

Follow step 3334,3336 or 3338, treatment scheme marches to step 3346.

In decision steps 3346, determine for the current ROI of the EDV(of ROI in expression for " validity " or the measurement of efficient on the specific increment observability of looking the unit) whether greater than predetermined value (for example, VALUE2).

If determine to be not more than predetermined value (VALUE2) for the EDV of ROI in decision steps 3346, then treatment scheme marches to step 3340.

If if determine in decision steps 3340 whether the border (and corresponding profile contour) of current region of interest can be simplified significantly (for example, using the method for Figure 30, wherein with the tolerance of ESO with the conservative simplification on the border that coaches).If ROI is DR _O, then the method for Figure 30 can directly be applied to this district.If ROI is DR _EOr the subarea, then the method for Figure 30 is applied to the district (SOR) that is blocked around current ROI.SOR can be restricted to the blocked area of next-door neighbour ROI.Alternatively, SOR can comprise other blocked areas that are connected to this SOR set.The method can expand to the ROI that adjacent region has been simplified fully with final realization so that conservative simplification is processed.

If determine that in decision steps 3340 border can be simplified, then in decision steps 3346, the EDV by the new conservative expression of blocking the district that the border surrounds of having simplified is determined.

On the other hand, can not be simplified the EDV value of realize target if determine the border of current ROI, then treatment scheme marches to step 3343.

In step 3343, current ROI is confirmed as having low EDV, thereby is used as for being left in the basket when the remarkable assembly of the increment observability of forward sight unit transition.In this step, if ROI is corresponding to DR _EAB, the mesh triangles shape that then is arranged in the correspondence in district conservatively is included in for the VM that looks unit A.In not by the new leg-of-mutton situation of the boundary segmentation introducing of DR, original triangle is included.If current ROI is corresponding to DR _OAB, then the mesh triangles shape of the correspondence in district conservatively is included in among the VM that looks unit B.Original triangle is included not introducing in the new leg-of-mutton situation by the boundary segmentation of DR.

If unified observability figure ROI is low corresponding to the EDV that the father and son looks unit transition and district, then the geometric figure around the district that is blocked conservatively is included among the ROI, and can recomputate the EDV of expansion area.For DR _ESituation, alternatively, can extend SOR district to adjacent district to exceed the SOR that tightly adjoins.The method can be identified following extreme case: father VM is not different significantly with sub-VM.Fully sub-ROI is removed in this case.

Step 3346,3340 and 3343 makes it possible to determine the ESV in district together, and if the value of ESV too low, increase ESV thereby then can attempt conservatively simplifying the border.If ESV still is lower than predetermined value, does not then think the corresponding obvious district with for the increment observability of looking the unit transition of ROI, and can ignore and look the unit transition.

Use ESV(to obtain according to ESO) reduced significantly the method required storage and computing time as the tolerance of the validity of ROI.Because in many cases, little blocking or expose the district can be owing to the again trigonometric ratio on ROI border is introduced a large amount of new triangles around ROI, so this sets up.These districts often have low ESO, thereby use this method it can not thought effectively to block (or exposing) district.Alternatively, for example, will simply conservatively think for the VM/PVS that looks unit A for recently as seen the collecting of primitive of specific AB transition.

If (for example, VALUE2), then treatment scheme marches to and processes 33-2, and it starts from step 3355 to determine to have surpassed predetermined value by the EDO of current ROI in decision steps 3346.

In step 3355, it can be to comprise the actual polygonal pointer that recently is blocked or the increment I-information of index information of referring to increment G+ and/or increment G-() or the son storage size of looking cell data (be father-son if look the unit transition) estimate, and the value of variable SS is arranged to become with this storage size certain direct ratio.The ROI that comprises many triangles/triangle segmentation often has for the increment G+ of correspondence or the high carrying cost of increment G-grouping.The clearly storage that the mark on the profile contour that limits corresponding ROI/VM border is replaced increment G grouping of the employed alternative storage format of method.Only when generating actual increment G information with the 3D trellis traversal of simplifying, the 3D trellis traversal of above-mentioned simplification adopts the border, unified VM district that generates according to for the concrete mark profile contour information of looking the unit transition.

Treatment scheme marches to step 3358, wherein value and the predetermined value of SS (for example, VALUE3) is compared.Be not more than VALUE3 if determine the value of SS in decision steps 3358, then treatment scheme marches to step 3361.

In step 3361, directly storage marches to decision steps 3388 for increment G data and the treatment scheme of ROI.

Decision steps 3388 is identical with previously described step 3349.

If determine the value of SS greater than predetermined value VALUE3 in decision steps 3358, then treatment scheme marches to step 3364.

Step 3364 is to 3385 being identification and the step of profile contour corresponding to the OB_SEG on the border that forms ROI (because ROI can comprise the hole, so be outer boundary and inner boundary).In these steps, the profile contour that mark is corresponding (it is that the limit of original triangle grid and summit add some additional sides corresponding to SE-MV wedge simple with inside corner and that the compound-contoured summit is located), and the man-to-man seed triangle of the coupling assembling of identification and storage and ROI.

From step 3364, to form with from looking OB_SEG that unit A looks and identifying from the outer boundary of looking ROI corresponding to OB_SEG that unit B looks and these OB_SEG of profile contour.

Treatment scheme marches to step 3367, and the OB_SEG that wherein surrounds ROI intersects each other and cross-shaped portion is appointed as IP and uses for the corresponding VM_INFO data structure (data structure that provides in Figure 31 A) of the profile contour of correspondence and store.

Treatment scheme marches to step 3370, wherein carries out mark for profile contour corresponding to the outer boundary of ROI and stores with the data structure (comprising alternatively DeltaGplus_attach_polyline info) of before describing in conjunction with Figure 31 A, Figure 31 B and Figure 31 C.

Treatment scheme marches to step 3373, wherein to form with from looking OB_SEG that unit A looks and identifying from the inner boundary of looking ROI corresponding to OB_SEG that unit B looks and these OB_SEG of profile contour.

Treatment scheme marches to step 3376, and the OB_SEG that wherein forms the inner boundary of ROI intersects each other and cross-shaped portion is designated as IP and use and store for the corresponding VM_INFO data structure (data structure that provides among Figure 31 A) of corresponding profile contour.

Treatment scheme marches to step 3379, wherein the profile contour corresponding with the inner boundary of ROI is carried out mark and stores with the data structure that previous geometric graph 31A, Figure 31 B and Figure 31 C describe.

Treatment scheme marches to from step 3379 and processes 33-4, and it is (Figure 33 D) from step 3382.In step 3382, to carry out mark and with the concrete unit transition phase of looking related for the concrete ROI that looks the unit transition corresponding all (may simplify) outline contours and interior profile contour.

The treatment scheme step 3385 of advancing wherein, will be stored in in concrete " the TRI_ seed _ tabulation " of looking the unit transition for a triangle of the assembly of each connection of ROI.

Then, if treatment scheme marches to step 3388 and there is not the not triangle of traversal in 3390(in VM).In some embodiments, these steps are identical with previously described step 3349 and step 3352 respectively.If there is the not triangle of traversal in VM, then treatment scheme marches to and processes 33-3, and it starts from step.

Looking ROI corresponding to unit transition with father-son is not the increment district from following meaning: look the unit and look as seen from look two of unit transition for father-son for the seed triangle of the ROI of the type.Use the ROI of the type, can with come according to the outer boundary of the mark profile contour structure of the correspondence of ROI and inner boundary according to father VM effectively constructor look the VM/PVS of unit.This structure uses 3D trellis traversal method hint, that simplify of Figure 36 and Figure 37.

By contrast, be observability increment district (DR) in abutting connection with looking ROI corresponding to transition between the unit.Use the ROI of the type, can be according to effectively constructing increment G+ and increment G-from looking unit A+B outer boundary and the inner boundary that mesh triangles shape/segmentation constructs together with the mark profile contour according to correspondence of ROI of looking visible.This structure also uses 3D trellis traversal method hint, that simplify of Figure 36 and Figure 37.

Figure 34 A and Figure 34 B: the mark profile contour information with the wall scroll contour is promptly constructed the method that observability figure blocks boundary segmentation in working time.

As previously described, identification is for (for example looking the unit transition, from A to B) effective increment district (DR), establish (method of using Figure 33) generate DR block the border correspondence the profile contour and come the above-mentioned profile contour of mark with the data structure of Figure 31 A and Figure 31 B.This mark is carried out as processed offline.

When storage mark profile contour information (such as the data that are associated with triangular mesh), can use this data based connection in working time or comprise the known observability figure structure observability figure corresponding with specifically looking the unit that looks the unit.

Figure 34 A be connected with Figure 34 B with previous storage for the mark profile contour information of specifically looking the unit transition according to connect or comprise and look known observability figure corresponding to unit and construct and the specific process flow diagram of looking the method for observability figure corresponding to unit.

In some embodiments, treatment scheme begins from step 3405, uses the labeling method that be used for to adopt Figure 30 and Figure 33 A and Figure 33 B that is associated with grid to come the data structure (Figure 31 A and Figure 31 B) of storage mark profile contour information to come identification marking profile contour (before specificly looking the unit transition generates and storage for what consider).

Treatment scheme marches to step 3410 and current stream shape triangular mesh (" grid ") is arranged to the concrete grid that the Contour.mesh field by the Contour data structure of Figure 31 refers to.

Treatment scheme marches to the concrete limit that step 3415 refers to the Contour.edge field of " current _ limit " (stream shape triangular mesh limit) being arranged to by the Contour data structure of Figure 31 A.

Treatment scheme marches to the concrete summit that step 3420 refers to the Contour.vertex field of " summit " (stream shape triangular mesh summit) being arranged to by the Contour data structure of Figure 31 A.

Treatment scheme marches to step 3425 so that variable " node counts " is arranged to 0.

Treatment scheme marches to step 3430, and wherein structure all observability figures corresponding with " current _ limit " block boundary segmentation (VM OB_SEG).With shown in Figure 35 and construct these VM OB_SEG in conjunction with the processing that this figure discusses in detail.

Treatment scheme marches to the concrete Contour_Node data structure that step 3435 refers to first node of variable " node " being arranged to refer to by current contour.

It is type 3 with the NODE.node_type that determines current C ontour_Node data structure (Figure 31 A) that treatment scheme marches to step 3450.If node_type is type 3, then node marches to processing 34-1 corresponding to compound-contoured summit and treatment scheme, and its step 3455 from Figure 34 B begins.

In step 3455, now " grid " variable (being initialised) is arranged to the CSV_INFO[NODE.NINFO that referred to by present node in step 3410] the concrete grid that refers to of .mesh, node_type is that 3 present node is compound-contoured summit (CSV) node.

Treatment scheme marches to step 3460, wherein variable " next one _ limit " (the next limit in the expression profile contour polyline) is arranged to the CSV_INFO[NODE.NINIFO of the CSV_Info data structure that referred to by current contour node] limit that refers to of .edge field.

Treatment scheme marches to step 3465 with the CSV_INFO[NODE.NINFO of the CSV_Info data structure variable " summit " being arranged to referred to by current contour node] summit that refers to of .point field.

Step 3455,3460 and 3465 effect are to locate two polylines on the face of one or more stream shape triangular mesh are linked together at a single point (compound-contoured summit).Be provided with current " grid ", " next one _ limit " and " summit " variable with this fusion that is reflected into the compound-contoured contour after, treatment scheme marches to processes 34-2, it is back to step 3480(Figure 34 A with treatment scheme).

If in decision steps 3450, determine " node _ type " be not 3(namely, this node does not correspond to the compound-contoured summit), then treatment scheme marches to step 3470.

In step 3470, with value initialization's one-tenth 0 of variable " limit counting ".

Treatment scheme marches to step 3475 variable " next one _ limit " is arranged to refer to the limit when front and (single order, look from looking the unit) silhouette edge of being connected to of stream shape triangular mesh.Can come easily this limit to be identified based on the connectivity of stream shape grid and the definition (referring to Fig. 3) of single order silhouette edge.Alternately, " next one _ limit " be identified as be connected to when limit in front, the limit that wherein connects has been marked as mark profile contour limit.This definition on " next one _ limit " is used in the following situation: the profile contour is fully skew or migration (because silhouette edge described above " retraction ") not.Situation be establish during pre-processing and be stored in the next_edge_type field of struct Contour data structure.If next_edge_type has 0 value, then " next one _ limit " be identified as previous mark the limit that connects of the next one.On the other hand, if the next_edge_type value is 1, then " next one _ limit " to be identified as be the limit that the next one of (single order) silhouette edge connects.

Treatment scheme marches to step 3480 so that " current _ limit " is arranged to " next one _ limit ".

Treatment scheme marches to step 3480, and wherein observability figure blocks boundary segmentation (VMOB_SEG) and produced by the cross-shaped portion of the umbra visibility events face that supported by current wedge with stream shape triangular mesh.These elements of deriving when the front from mark profile contour of observability figure are to use at the process flow diagram of Figure 35 and in conjunction with the method that this figure discusses in detail to construct.

(use in the method shown in the process flow diagram of Figure 35) when constructing the VM OB_SEG that is generated by current silhouette edge, treatment scheme marches to step 3485.

In step 3485, increase variable " limit counting ".

Treatment scheme marches to step 3488 to determine that " limit counting " is whether less than the span_length(NODE.span_length for present node of appointment in the Contour_Node of Figure 31 A data structure).

If determine that in decision steps 3488 " limit counting " less than span_length, then treatment scheme is back to step 3475, wherein identifies next limit.

On the other hand, be not less than span_length if determine " limit counting " in decision steps 3488, then treatment scheme marches to decision steps 3490.

Determine the number of nodes that provided by data field CONTOUR.num_nodes during " node counts " is whether less than contour in decision steps 3490, wherein CONTOUR is to just referring to and adopt the data structure Contour shown in Figure 31 A at processed current mark profile contour.

If determine " node counts " less than CONTOUR.num_nodes in decision steps 3490, then treatment scheme marches to step 3492.

In step 3492, increase " node counts " and processing are back to step 3435, wherein select next node and process continuation.

On the other hand, be not less than CONTOUR.num_nodes if determine " node counts " in decision steps 3490, then treatment scheme marches to step 3495.

Step 3495 expression is treated, and mark profile contour and processing should march to next mark profile contour.Treatment scheme stops at step 3495 place.

Come carrying out overall control for the specific processing of looking all mark profile contours of unit by processing shown in Figure 36 and that discuss in detail in conjunction with this figure.

Figure 35 A and Figure 35 B: the observability figure that showing structure derives from the wall scroll silhouette edge of mark profile contour blocks the process flow diagram of the method for boundary segmentation.

Figure 34 is the process flow diagram that is used for each limit of identification marking profile contour in the situation of data structure for the low volume data of whole contour storage of given use Figure 31 A.Figure 35 is that structure is from the process flow diagram of the processing of the VM OB_SEG of the single edge derivation of mark profile contour.

By the processing of calling Figure 35 during the processing (step 3482 of Figure 34) to the profile contour, the processing of Figure 34 has caused the structure for all VM_OBSEGD of whole mark profile contour effectively.

In some embodiments, when running into the silhouette edge of mark, treatment scheme is from step 3510.This is corresponding to the step 3482 of the processing of calling Figure 34.

Treatment scheme marches to decision steps 3520 with the span_type that determines current mark profile contour node (" node " transmits from calling to process) and appointment the specified Contour_Node.node_type data field of Figure 31 value of equaling 1 whether.

Form external corners if in decision steps 3520, determine the mark profile contour that the span_type value of equaling 1(of current mark profile contour node is illustrated in the Contour_Node.node_type data field that uses Figure 31 A), then treatment scheme marches to step 3450.

In step 3540, process to construct the SV-ME wedge with the rotation of before describing in conjunction with Fig. 4.

Treatment scheme marches to step 3580 with the current silhouette edge of determining the to have constructed wedge article one limit in the contour whether.

If determine that in decision steps 3580 silhouette edge is article one limit in the contour, then treatment scheme marches to step 3592.

In step 3592, the wedge that obtains pre-stored according to CONTOUR.VMinfo.point, CONTOUR.VMinfo.mesh and the CONTOUR.VMinfo.tri data field of appointment in the data structure of Figure 31 A-mesh triangles shape is intersected information.This is corresponding to the cross-shaped portion of wedge and the leg-of-mutton precomputation of specifiable lattice and pre-stored, and wherein this cross-shaped portion is the initial OB_SEG that comprises the VM OB_SEG chain on the VM border that is associated with mark profile contour.Identification for the specific profile contour of looking the unit transition, produce have the minimum effectively dynamically processed offline of the value of blocking during in advance precomputation and store this information (step 3381 of Figure 33).Therefore, if this limit is article one limit in the mark profile contour, then step 3592 generates corresponding VM border the one OB_SEG.

On the other hand, if determine that in decision steps 3580 silhouette edge of processing is not article one limit in the contour, then treatment scheme marches to step 3585.

In decision steps 3585, determine whether present node has CSV_Info associated with it, that is, whether current silhouette edge supports the wedge that intersects with another visual profile limit.If so, then the Contour_Node.node_type value can equal 3(Figure 31 A) and the Contour_Node.ninfo value can be index in the CSV_Info array of data structures (Figure 31 B).In one embodiment, without mark profile contour storage CSV_Info data but alternatively the initial point with every contour is limited to the CSV place, thereby the data of correspondence are stored in Contour.mesh, Contour.triangle and the Contour.vertex field.

If determine that in decision steps 3585 present node is associated with the CSV_Info data, then treatment scheme marches to step 3590.

In step 3590, from CSV_Info[Contour_Node.inifo] .mesh, CSV_Info[Contour_Node.inifo] .edge and CSV_Info[Contour_Node.inifo] read initial VMOB_SEG data the .point data structure.

On the other hand, if determine that in decision steps 3585 present node is not associated with CSV_Info, then treatment scheme marches to step 3595.

In step 3595, use VM OB_SEG grid, triangle and some intersection data from last silhouette edge in the contour to construct the VM OB_SEG corresponding with current silhouette edge.Because VM OB_SEG forms the polyline on the face of stream shape triangular mesh, so be the direct segmented construction of the polyline on the polyhedron from the structure of the OB_SEG of adjacent edge.

Follow step 3590 or step 3595, treatment scheme marches to be processed 35-1 and processes 35-2, and it is from the decision steps 3596 of Figure 35 B.In decision steps 3596, determine whether the VM_INFO.RETRIANGULATE_HINT field is not equal to 0.This is the field of the VM_INFO data structure of Figure 31 A, and if determine to adjoin effectively dynamically blocking and can increasing by the again trigonometric ratio that blocks boundary of observability map-area, then between the working time of observability figure tectonic epochs with this field value of being arranged to 1.

Be not equal to 0 if determine the value of VM_INFO.RETRIANGULATION_HINT in decision steps 3596, then treatment scheme marches to step 3597.

In step 3597, carry out again trigonometric ratio at boundary to adjoining the corresponding triangle that blocks the border.

On the other hand, if determine that in decision steps 3596 VM_INFO.RETRIANGULATION_HINT equals 0, then treatment scheme marches to step 3599.

In step 3599, do not carry out again trigonometric ratio at boundary to adjoining the corresponding triangle that blocks the border.The triangle " link " that in this case, will comprise the profile contour does not carry out again trigonometric ratio to partly being blocked triangle.Treatment scheme stops at step 3599 place.

If determine that in decision steps 3520 NODE.span_type is not equal to value 1, then treatment scheme marches to step 3525.

Determine whether value of equaling 2 of NODE.span_type in decision steps 3525, if the value of equaling 2, then treatment scheme marches to step 3545.If node_type value of equaling 2, then contour can comprise external corners profile summit and inside corner profile summit.Thereby, treatment scheme sequentially proceed to 3545 with determine current silhouette edge whether relate to the external corners or inside corner of next bar silhouette edge of contour in.

If that determines the profile contour in decision steps 3545 forms external corners when front and next bar limit, then treatment scheme marches to step 3540, as previously described.

On the other hand, if determine the formation inside corner when front and next bar limit of profile contour in decision steps 3545, then treatment scheme marches to step 3555.

In step 3555, form the SE-MV wedge that is incident on the inside corner profile summit with the before processing of scanning in conjunction with Fig. 5 A and Fig. 5 B description.Thereby treatment scheme marches to step 3580, as previously mentioned.

If determine that in decision steps 3525 NODE.span_type is not equal to value 2, then treatment scheme marches to step 3530.

Whether the value of determining NODE.span_type in decision steps 3530 equals 3.Equal 3 if determine the value of NODE.span_type in decision steps 3530, then treatment scheme marches to step 3560.In this case, span_type represents and simplify contour by the subset of skipping the limit of contour between umbra wedge/VMOB_SEG tectonic epochs.Such as the effective static barriers of difference (be also referred to as effectively dynamically and block) of the identification DR in the step 3075 and 3089 of Figure 30 B with during simplifying the profile contour this information is carried out precomputation and this information is pre-stored among the corresponding Contour_Node.span_type and Contour_Node.span_length data structure.

In step 3560, NODE.span_length and variable SPAN_COUNTER(were initialized to 0 in the past running into contour) compare with determine should those summits at the profile contour between the umbra visibility events face simplified of structure and corresponding VM OB_SEG.If determine SPAN_COUNTER less than NODE.span_length in decision steps 3560, then treatment scheme marches to step 3565, and its expression is not forming wedge when the front.

Then treatment scheme marches to step 3570, wherein current silhouette edge is directly linked to the single OB_SEG for whole profile contour, and it is when decision steps 3560 guidance of faulf handlings final structure in step 3540 or 3545 during towards step 3545 or step 3540.

If determine that in decision steps 3530 NODE.span_type is not equal to value 3, then treatment scheme marches to step 3535.

In decision steps 3535, determine whether value of equaling 3 of NODE.span_type.

If determine NODE.span_type value of equaling 3 in decision steps 3535, then treatment scheme marches to step 3575.In this case, span_type represents not should be with scanning the umbra event face of constructing on the inside corner profile summit that is incident on contour of processing, but should be with the better simply method of the Plane intersects of adjacent S V-ME wedge is constructed.

Thereby, in step 3575, with adjacent S V-ME wedge the plane intersect construct SE-MV wedge (and corresponding VM OB_SEG) and treatment scheme marches to previous described step 3580.

Figure 36 shows the process flow diagram of the processing that the runtime processing of using the ROI border of constructing according to the mark profile contour of pre-stored to construct observability figure ROI is controlled, and wherein the ROI border limits/delimited simplification, hint, 3D trellis traversal processing working time that ROI is traveled through.

As before described in conjunction with Figure 33 A to Figure 33 D, can will look look visible mesh triangles shape segmentation and comprise with two and look the region of interest of blocking the border (ROI) from looking to look the unit corresponding to unit of unit for comprising among the unified observability figure from two for specific increment observability data description of looking the unit transition.

The type of increment observability data depends on the structure of corresponding ROI, and it depends on the relation that the unit transition is looked in two descriptions of looking the unit.

Look in the unit if be comprised in another fully depending on one of unit, then ROI can be corresponding to only from the involved district that looks visible, unit of looking.The mark profile contour corresponding with the ROI of these types can be used for looking the VM(that the VM of unit comes constructor to look the unit in working time according to the father and use 3D trellis traversal hint, that simplify), thus clearly storage to increment G-information avoided under specific circumstances.

If look the unit and share face and one and look the unit and be not comprised in another and look (relation that is called " adjacency ") in the unit for two, then ROI is corresponding to increment district (DR).For looking unit transition (being called the AB transition) from VC A to VC B, DR has two types.One type increment district DR _OAB(blocks the increment district from A to B) comprise from looking unit A look visible from looking the unit B mesh triangles shape segmentation of looking sightless.DR _OAB also is DR _EBA(exposes the increment district from B to A).Similarly, DR _EAB=DR _OBA.DR _OAB is corresponding to for the increment G-data of AB transition and DR _EAB is corresponding to the increment G+ data for the AB transition.Mark profile contour corresponding to the border of these ROI can be used in operation is structure increment G+ and/or increment G-data (also using 3D trellis traversal hint, that simplify), thereby avoids under specific circumstances the clearly storage to increment G+ and increment G-polygon segment information.

In some embodiments, treatment scheme begins from step 3605, wherein will " LAB_CON_ tabulation " visits as array for the tabulation of all mark profile contours of specifically looking the unit transition.

Treatment scheme marches to step 3610, and wherein each in the contour of the mark in " LAB_CON_ tabulation " stands further processing.

In first step that the mark profile contour in " LAB_CON_ tabulation " is processed, treatment scheme marches to step 3615, wherein come the limit of identification marking profile contour with the processing shown in the process flow diagram of Figure 34, and above-mentioned limit is stored in " LAB_SIL_ limit _ tabulation ".

Treatment scheme marches to step 3620, and wherein further processing is stood on the limit on " LAB_SIL_ limit _ tabulation ".

In the first step that the limit in " LAB_SIL_ limit _ tabulation " is processed, treatment scheme marches to step 3625, wherein constructs the VM OB_SEG corresponding with the limit on " LAB_SIL_ limit _ tabulation " with the processing shown in the process flow diagram of Figure 35.

Treatment scheme marches to decision steps 3630 to determine that whether " LAB_SIL_ limit _ tabulation " be as empty.If in " LAB_SIL_ limit _ tabulation ", have more limit to process, then select next untreated limit and and process and to be back to step 3620.

If determining in decision steps 3630 does not have more limit to process in " LAB_SIL_ limit _ tabulation ", then treatment scheme marches to step 3635.

In decision steps 3635, determine in " LAB_CON_ tabulation ", whether to have the contour of arbitrarily more mark to process.If determining in decision steps 3635 has the contour of more mark to process in " LAB_CON_ tabulation ", then select next untreated contour and treatment scheme to be back to step 3610.

On the other hand, if determine do not have the contour of more mark to process in " LAB_CON_ tabulation " in decision steps 3635, then treatment scheme marches to step 3640.

In step 3640, will be as the array of being arranged to be called " TRI_ seed _ tabulation " for the triangle seed list (precomputation and pre-stored in the step 3383 in Figure 33 B) of the tabulation of the leg-of-mutton precomputation that refers in each increment district corresponding with specifically looking the unit transition.In this case, the triangle seed list comprise come from look unit A(comprise look the unit) looking is not blocked and from looking the look triangle in each VM district of not being blocked of unit B.From VM A from looking unit A and looking unit B each the VM district that looks visible and select a seed triangle and so that initiate the traversal of the leg-of-mutton set of seed is guaranteed to run into relevant (mark) profile contour for the AB transition during the 3D trellis traversal of simplifying the working time of Figure 37.Leg-of-mutton this selection of seed is guaranteed to construct the VM that looks unit B with following traversal processing according to the VM that looks unit A: usage flag profile contour information " is avoided " around geometric figure in the AB transition.

Treatment scheme marches to step 3645, and wherein the triangle of " TRI_ seed _ tabulation " is through being subject to processing.

Treatment scheme marches to step 3650, wherein uses the triangle of " TRI_ seed _ tabulation " to initiate such as the stream shape trellis traversal in the simplification of at length discussing shown in the process flow diagram of Figure 37 and in conjunction with this figure.This traversal is by to identifying from looking unit B all triangles of looking visible from looking the look traversal of little subset of visible triangle (those triangles " TRI_ seed _ tabulation ") of unit A.

Treatment scheme marches to decision steps 3660 to determine in " TRI_ seed _ tabulation " whether any how untreated triangle being arranged.

If determining in decision steps 3660 has untreated triangle in " TRI_ seed _ tabulation ", then select next untreated triangle and treatment scheme in " TRI_ seed _ tabulation " to be back to step 3645.

On the other hand, if determine there is not how untreated triangle in " TRI_ seed _ tabulation " in decision steps 3660, then treatment scheme marches to step 3665.

Step 3665 expression has been constructed the specific ROI corresponding with required increment visibility information by traversal working time of simplifying.As previously discussed, when the unit transition of looking of correspondence is father when arriving the transition of son, in fact this increment visibility information can be to look the look set of visible triangle/segmentation of unit from son.Alternately, this ROI information can be corresponding to the increment G+ that looks the unit transition and the increment G-data between the unit of looking for adjacency, thereby so that can select when needed to generate increment G grouping rather than storage for all increment G groupings of looking the unit transition at every turn.Treatment scheme stops at step 3655 place.

Figure 37 processes main processing of tabulating for the seed triangle of the assembly of the connection of this ROI according to mark profile contour information structuring ROI and the structure of pre-stored with 3D trellis traversal working time of simplification, hint.

Figure 37 show being instructed of in the step 3650 that the control shown in Figure 36 is processed, calling working time the 3D traversal processing the process flow diagram of embodiment.

The general 3D trellis traversal of processing shown in the process flow diagram of Figure 37 and Figure 20 A is processed quite similar.The processing of Figure 20 A generally as off-line pre-service carry out being in precomputation and store PVS and the purpose of mark profile contour data is constructed observability figure.

By contrast, to process be to carry out and be used for generating for the specific ROI that looks unified the observability figure of unit transition in working time to the 3D trellis traversal of Figure 37.These ROI are described the variation of the observability that causes because these specifically look the unit transition.

One type look in the unit transition, the uncle looks the unit and looks the unit to involved son, and ROI only comprises from son and looks unit those the triangle/segmentations of looking visible.When the seed triangle that provides is to comprise self-contained when looking the unit and look visible the leg-of-mutton triangle set of assembly of each connection of leg-of-mutton those ROI from son, can process to generate the ROI of the type according to 3D trellis traversal simplification, hint of Figure 37.

The relevant ROI border of using these seed triangles and generating according to mark profile contour, so that 3D trellis traversal processing that simplify, hint walks around or " avoiding " (for example specifically looking the unit transition period, from looking unit A to looking unit B, wherein look unit B be comprised in look in the unit A) polygon or the polygon segmentation that become and recently be blocked.This makes it possible to remove polygon and/or the polygon segmentation that recently is blocked in the situation of the polygon tabulation that ambiguously storage will remove.If the polygon that removes tabulation is very large with respect to the quantity of visible polygon, then to compare and use the clear and definite polygonal tabulation that will remove, this can be more effective.

Can also divide into groups both with traversal method the increment G-that looks the unit transition and the increment G+ that look between the unit that directly generate for two adjacency of hint, one of them look the unit be not comprised in another look in the unit but two look the unit and have shared face.In this case, the VM of beginning must comprise from combination and looks unit A+B look visible all polygons or polygon segmentation.In addition, beginning VM must comprise for looking unit A and the relevant border (these block the border all is that silhouette edge according to mark generates) of blocking of looking unit B.Such observability figure is called unified observability figure.Comprise for the unified observability figure that looks the unit of two connections and to look visible and from looking look all polygon segmentations of visible (or look unit A+B from Boolean add and look as seen) of unit B from looking unit A.In addition, unify observability figure comprise with look unit A and look unit B corresponding from looking the border of blocking that the unit looks.

In order to generate for A to the increment G+ of the transition (being also referred to as the AB transition) of B grouping, use to be blocked for looking from A and to come the unified VM of correspondence is traveled through from the look seed triangle of assembly of each connection in visible VM district of B.The district of the type is called DREAB.Use these seed triangles initiate traversal and traversal march to look unit A or look unit B corresponding block the border.Run into and look A border, unit in the side that is blocked depending on A border, unit and run into exposing side.This corresponding in Figure 40 with the DReAB district shown in the light gray.

Table X IV has summed up and has initiated to be used for to generate in abutting connection with the required leg-of-mutton set of seed of traversal working time of the hint of the increment G+ that looks the unit and increment G-grouping and show the side of blocking the border that runs into.

The method makes it possible to basis for the unified VM that looks the unit transition and corresponding VM blocks border contour (it can generate according to mark profile contour) generation increment G+ and increment G-divides into groups both.Use the method, need to not store clearly for the increment G+ that looks the unit transition and increment G-grouping at every turn for look the unit transition at every turn.More specifically, can any time before the needs grouping become to divide into groups next life by the traversal method that hints.

In addition, in distributed clients-server implementation, generate the increment visibility information working time the 3D traversal method can be more effective.In the district of the high visible correlativity of space-time, can specifically look the contour information that the unit uses identical mark continually for some in the same zone.Therefore, can calculate the observability figure of increment/PVS with the required less the transmission of data of the transmission of data of direct polygon increment G-tabulation that will remove to compare for look the unit transition at every turn to 3D trellis traversal working time based on the profile information of mark.

Now turn to Figure 37, in first step that travels through the working time of simplifying, treatment scheme is wherein initiated traversal for specifically looking the specific triangles place of unit transition in " TRI_ seed _ tabulation " from step 3705.

Whether treatment scheme marches to decision steps 3715 and comprises from looking the border of blocking that the unit looks with the triangle of determining to be traveled through.Can in the step 3620 of Figure 36, construct these borders.

Comprise from looking the border of blocking that the unit looks if determine the triangle travel through in decision steps 3715, then treatment scheme marches to step 3735.

Treatment scheme marches to step 3735, is wherein blocking boundary interruption traversal.In other embodiment, treatment scheme marches to from 3735 and processes 37-1, and it is back to step 3725 with treatment scheme.

On the other hand, if determining in decision steps 3715 that the current triangle that travels through does not comprise blocks the border, then treatment scheme marches to step 3720.

In decision steps 3720, determine the current triangle that travels through whether have with for silhouette edge corresponding to the silhouette edge of the specific mark of looking the unit transition of being considered.The silhouette edge of these marks is corresponding to the inner boundary of the corresponding ROI of unified VM.

If determine in decision steps 3720 whether the current triangle that travels through has silhouette edge, and then treatment scheme marches to step 3725.

In step 3725, select next (connection) do not travel through in the grid triangle and with its be committed to step 3715 and subsequent step with for the treatment of.

On the other hand, if determine that in decision steps 3720 current triangle does not comprise the silhouette edge for the mark of working as the transition of forward sight unit, then treatment scheme marches to step 3780.

(not traversal) triangle of in decision steps 3780, determining in current " traversal ", whether there be any being untreated, wherein herein traversal is defined as by cold silhouette edge connect and be arranged in as step 3625 structure of Figure 36 for the polygonal set of grid on the specific side that is not blocked of blocking the border of looking the unit transition.

If determining in decision steps 3780 has the not triangle of traversal in current traversal, then treatment scheme marches to step 3725, wherein selects the next triangle of traversal to be used for as the aforementioned data.

On the other hand, if determine in current traversal not the not triangle of traversal in decision steps 3780, then treatment scheme marches to step 3788.

In step 3788, the side that is not blocked of blocking boundary segmentation of constructing in the step 3625 of Figure 36 continues traversal; The wherein said border of blocking is corresponding to the silhouette edge of the mark that runs in step 3720.This continuation can relate to and continued traversal by the triangle of trigonometric ratio again (that is, block boundary accurately pruned), maybe can relate to not at boundary being continued traversal by the leg-of-mutton side that is not blocked of trigonometric ratio again.Treatment scheme stops at step 3788 place.

Figure 38: will with the method that is attached to the beginning border of corresponding mark profile contour for increment G+ sub-grid corresponding to the specific grid elements of recently exposing of looking the unit transition.

Look the unit transition for any, can be at the polyline that structure is determined during the processed offline (Figure 33) of looking the observability figure that looks the unit and corresponding observability increment district and storage is corresponding with the attached curve of the sub-grid that recently exposes.Can store these polylines with the DeltaGplussummesh_attach_polyline data structure of Figure 31 B.Can be stored in these data and during the runtime processing of Figure 36 and the described increment observability figure of Figure 37/PVS structure, use these data subsequently.Particularly, in order to use during working time 3D trellis traversal method working time of determining the grid elements that recently is blocked by " avoiding ", at attached polyline place any grid elements of recently exposing is connected to the element that had before exposed and travels through working time to guarantee that the stream shape that will connect is presented to.

Note, in some embodiments, when the method with Figure 36 and Figure 37 is used for using the method for Figure 38 by getting around when calculating the geometric figure that recently is blocked working time.

In some embodiments, treatment scheme marches to step 3805, wherein conducts interviews for the DeltaGplussummesh_attach_polyline data structure that is associated for the specific silhouette edge (beginning) of looking the mark of unit transition and refers to this data structure with variables A P.

Treatment scheme marches to decision steps 3810, determines that wherein whether attached type is corresponding to 0.If determine attached type corresponding to 0 in decision steps 3810, then treatment scheme marches to step 3815.

In step 3815, the limit of increment G+ sub-grid (herein referred to as sub-grid) is connected directly to the corresponding sides of main grid.The free margins (only having a polygonal limit of assembly) that the single limit of listing from AP.submesh_edgelist and AP.mainmesh_edgelist begins along the grid of correspondence is traveling between AP.edgenumber the limit and makes this connection.Under this pattern, only need the single limit of pre-stored for each edgelist.

Treatment scheme marches to decision steps 3820 to determine that whether attached type is corresponding to 1.If determine attached type corresponding to 1 in decision steps 3820, then treatment scheme marches to step 3825.

In step 3825, the limit of increment G+ sub-grid (herein referred to as sub-grid) is connected directly to the limit of the correspondence of main grid.The first side of listing from AP.submesh_edgelist and AP.mainmesh_edgelist begins sequentially to run through the tabulation of whole limit and make this connection between AP.edgenumber limit.

Treatment scheme marches to decision steps 3830 to determine that whether attached type is corresponding to 2.If determine that in decision steps 3830 attached type is to equal 2, then treatment scheme marches to step 3835.

In step 3835, with the limit " link " of increment G+ sub-grid (herein referred to as the sub-grid) corresponding sides to main grid, wherein link can be from an ambiguity line segment to another polyline one to many mappings.During pre-processing these links are identified so that they present the conservative expression of corresponding grid during the 3D traversal in working time.The free margins (only having a polygonal limit of assembly) that the single limit of listing from AP.submesh_edgelist and AP.mainmesh_edgelist begins along corresponding grid is traveling between AP.edgenumberge the limit and makes this link.Under this pattern, only need storage and the single limit of storage for each edgelist.

Treatment scheme marches to decision steps 3840 to determine that whether attached type is as 3.If determine attached type corresponding to 3 in decision steps 3830, then treatment scheme marches to step 3845.

In step 3845, with the limit to the correspondence of main grid, the limit " link " of increment G+ sub-grid (herein referred to as sub-grid), wherein link can be from an ambiguity line segment to another polyline one to many mappings.During pre-processing these links being identified makes them present the conservative expression of corresponding grid during the 3D traversal in working time.The first side of listing from AP.submesh_edgelist and AP.mainmesh_edgelist begins and sequentially runs through the tabulation of whole limit to be traveling between AP.edgenumber the limit and to make this link.

Under any circumstance, this looks the unit transition and marches to step 3850 treatment scheme for next.In step 3850, can derive according to the end boundary when the transition of forward sight unit the next beginning border of looking elementary boundary, and processing marches to decision steps 3810 for look the unit transition next time.Treatment scheme stops at step 3850 place.

In some embodiments, except using for the polygonal increment G+ geometric figure grouping of recently exposing, do not needing in the clear and definite polygonal situation of increment G+, at some faces that recently expose of program ground reconstruct working time.In some embodiments, adopt the method in the increment district of recently exposing, to generate for example recently visible part of floor or ceiling of single large polygon (or chessboard trellis face).In this case, the mark of particularly instruction runtime processing program ground part of recently exposing of generation face in whole increment district comes mark profile contour.

Figure 42 A and Figure 42 B show and use level to look the process flow diagram of the method for unit.

According to before coming Incremental Construction to make it possible to determine the triangle that recently is blocked by the district that recently is blocked that during simplifying, effectively walks around or avoid observability figure from the method (Figure 34 to Figure 38) of the observability figure that looks the unit and look from looking the observability figure that looks the unit.The method makes it possible to do not removing the geometric figure that recently is blocked in storage or the situation of transmission for the leg-of-mutton information that recently is blocked particularly.Alternatively, use to construct for the profile contour information that is labeled of dynamically blocking profile contour (DOSC) and walk around the geometric new observability figure that is blocked and block the border.

Use the mark profile contour of the data structure coding of Figure 31 A and Figure 31 B often to have the storage demand in low-down increment district for the dynamic observability correlativity of height, this mainly is because can reuse identical contour many times looking the unit transition period.Thereby, compare for every this look the unit transition must storage/transmission for the increment PVS scheme of the routine of each leg-of-mutton information that recently is blocked, increment observability figure maintaining method can have fully less storage/transmission cost.

This especially sets up when adopting following methods: use the hierarchical organization of looking the unit, some of them look unit and other look the unit adjacent and other look the unit and be completely contained in other and look in the unit (that is, they are that the son that the father looks the unit is looked the unit).

Use this to look cell layout, be configured to the large observability figure that looks the unit according to the adjacent observability figure that looks greatly the unit; And the observability figure that looks the unit according to the father comes constructor to look the observability figure of unit.Under latter event, can not come constructor to look the observability figure of unit with increment G+ information, this is because look cell moving when looking the unit to involved son as the uncle, does not have geometric figure can become recently as seen.With son look observability figure corresponding to unit according to the father look the unit observability figure Incremental Construction only needs determine the geometric figure recently be blocked, should determine to use increment observability figure structure working time (Figure 34 to Figure 38) to realize, it only uses the DOSC data of the mark that the triangle that causes recently being blocked " avoids ".

Can also improve server from supply increment G+ and increment G-information to the data access of playing up client and transfer efficiency to the larger adjacent use of looking the unit.This is because disk access/searching number of times is often arranged totally reading the time of little packet, thereby so that the transmission of larger packet is more effective.

In some embodiments, treatment scheme determines that in step 4210 whether the unit transition of looking of (method of using Figure 33) the increment VM/ increment PVS information of will encoding looks the unit and (involved) son is looked between the unit the father from step 4210.

Look the unit is looked in the unit to son transition if determine to look the unit transition in decision steps 4210 corresponding to the uncle, then treatment scheme marches to step 4230.

In step 4230, the increment VM/ increment PVS information corresponding with looking the unit transition (method with Figure 33 and relevant figure is determined) is corresponding to DOSC mark profile contour information.

If determine to look the unit transition corresponding to not being that the uncle looks the unit is looked in the unit to son transition in decision steps 4210, then treatment scheme marches to step 4220.

In step 4220, the increment VM/ increment PVS information corresponding with looking the unit transition (method with Figure 33 and relevant figure is determined) is corresponding to DOSC, DESC and increment G+ sub-grid information.Treatment scheme stops in step 4220 and 4230 places.

Embodiment adopts larger father to look the unit and calculates farther geometric observability and adopt less son to look the unit and calculate nearer geometric observability.Because along with the observability of viewpoint motion changes (for example, recently to expose and the face speed of development that recently is blocked) generally along with diminishing apart from the increase of the distance of looking the unit, so the method is effective.

Now turn to Figure 42 B, whether treatment scheme is very little from the size that step 4240 begins to determine to look with respect to the distance of determining the particular geometric figure of dynamic observability (for the increment VM/ increment PVS data of looking the unit transition) unit.

If it is very little to determine to look with respect to geometric distance the size of unit in decision steps 4240, then treatment scheme marches to step 4260.

In step 4260, be not included in for the increment VM/ increment PVS calculating of looking the unit transition that relates to when the forward sight unit apart from the predetermined particular geometric figure than distant location of looking the unit.This geometric figure can be included in for the father who relates to when the forward sight unit and look in the increment VM/ increment PVS calculating of looking the unit transition of unit.

If determine that in decision steps 4240 it is not very little looking the unit with respect to geometric distance, then treatment scheme marches to step 4250.

In step 4250, be included in for relating in the increment VM/ increment PVS that looks the unit transition of forward sight unit calculates apart from the predetermined particular geometric figure than distant location of looking the unit.Treatment scheme stops in step 4250 and 4260 places.

Figure 43 A and Figure 43 B show the data structure of safeguarding for the increment VM/PVS that uses increment VM/PVS data.

Figure 43 A and Figure 43 B show the data structure of the method for safeguarding for the increment observability figure PVS that processes by client and server to carry out.No matter client and server is processed on the different physical equipment that whether resides on by network connection, can use data structure herein.Can be alternatively, can carry out at identical computing equipment adopting these data structures in the enforcement of two processing.

DDL is purpose display list (also can be called dynamic display list).This tabulation can be comprised of one or more tabulations or the array of the data structure that represents the polygonal mesh object.These comprise triangle (DDL_Tri), limit (DDL_Edge), summit (DDL_Vertex) and potentially other data that are associated such as material and texture description etc.Other data structures that are used for enforcement DDL method can find in the Table X III in the inventor's the U.S. Patent No. 6,111,582.

In U.S. Patent No. 6,111, in 582 the method, the DDL data structure is used for putting maintenance into practice for every image DDL scheme of looking from viewpoint of the dynamic display list of single image.By contrast, for the present invention, adopt the DDL method safeguard with the current view point that is located substantially on the user near DDL set corresponding to the conservative PVS that looks the unit of (or being arranged in the expansion area that is connected near district this).Yet can in two kinds of methods, adopt some data structures that comprise about the data structure of material, texture, procedural texture and other etc.

So that the DDL set can be used for client process.DDL and should near the PVS set pair as being determined by the navigation prediction algorithm of looking the unit the viewpoint.Struct PVS data structure with Figure 43 B refers to DDL.

Figure 44 A, Figure 44 B, Figure 44 C with used Global_to_DDL_Linkmap_Edge, the Global_to_DDL_Linkmap_Tri shown in Figure 43 A, Global_to_DDL_Linkmap_Vertex data structure shown in the relevant figure and in conjunction with the method that those figure discuss in detail.

Figure 44 A, Figure 44 B and Figure 44 C show and support to use the increment VM/PVS(increment G+ sub-grid that sends from remote server) the increment VM/PVS of data data storage and the transmission method safeguarded.

In one embodiment of the invention, all data (for example are stored in permanent, non-nuclear storer, disk), the increment G+ sub-grid, VM_Info, DeltaGplussubmesh_attach_polyline, Contour and other data structures that comprise Figure 31 A and Figure 31 B refer to mesh triangles shape, limit and summit so that the element of storing refers to symbol with overall ID.Global ID is for unique integer on each element (for example, single triangle, limit, summit) in the whole database of storing of the complete 3D model of expression.Global ID is the index in the element array (a plurality of grids, triangle, limit and summit) for whole model.Leg-of-mutton Global ID is for the triangle index in the overall triangle array of whole model.The Global ID on limit is for the limit index in the overall limit array of whole model.Similarly, the Global ID on summit is for the vertex index in the overall vertical array in the whole model.

The method that the process flow diagram of Figure 44 A to Figure 44 C provides makes it possible to be undertaken playing up working time by following client process/machine: its any one the time be engraved in the little subset of only storing whole graphic data base in the data structure that is called purpose display list (DDL).DDL comprises the pictorial element (geometric figure and texture information) that is referred to by current all VM/PVS for looking the unit that are stored on the client.By carry out the method for Figure 44 A to Figure 44 C for the server process of all increment G+ sub-grid data that read by server process.The method of Figure 44 A to Figure 44 C has two major functions: 1) Global ID is referred to symbol and convert direct DDL to and refer to symbol, 2) prevent to the data transmission of client DDL redundant.

Now provide the general description of method, then the process flow diagram based on Figure 44 A, Figure 44 B and Figure 44 C comes the method is described in detail.

Server process is fetched in advance from client viewpoint position data and presumptive navigation and is read and specific increment VM/PVS data of looking pre-stored corresponding to unit transition.Server maintenance is used by server process and is sent to the increment VM/PVS grouping of client process at the current VM/PVS data acquisition of client process maintenance by the DDL set that current VM/PVS data acquisition refers to.Before increment VM/PVS data were sent to client process, server process converted all Global IDs to DDL direct index or other local I D that refers to symbol.In addition, server process checks whether each increment G+ sub-grid grouping of accessing has been presented among the corresponding DDL with triangle or other graphic elements in definite increment G+ sub-grid grouping with DDL.If this element has been presented among the corresponding DDL, then do not need this element is added into DDL.

If server process is positioned on the physically different computing equipment (for example, connecting by the grid agreement) with client process, then client DDL is the mirror image by the DDL of server maintenance.In this case, element in the grouping of server inspection increment G+ sub-grid can be eliminated from the redundant transmission of server to client end to determine the preceding method whether they appear among the DDL, thereby has reduced fully the required connection bandwidth of latest copy of server maintenance DDL.

For the embodiment that carries out the client and server processing in identical physical computing devices (shared processing device, internal memory and external memory), can safeguard single part of DDL copy.In this case, because owing to the identical surrounding area that the unit often is positioned at model of looking of interior storage causes individual element (for example, triangle) probably appears among the current some VM/PVS for looking the unit that are stored in the internal memory, so DDL has reduced the memory cost.Therefore, store independently each element if compare for each VM/PVS, store the graphic element of being shared by some VM/PMS data structures with DDL and generally have lower carrying cost.

For crossing the processing that server process is carried out Figure 44 A to Figure 44 C from all increment G+ sub-grid data communication devices of external memory access looking the unit transition period.Usually, client process is by controlling the viewpoint navigation via mouse, gamepad controller etc. from user's real-time input.Server process receives about the data of the current location of viewpoint and adopts these data to use and comprises navigating around the supposition of looking the unit of dead reckoning and the navigation Forecasting Methodology of colliding constraint.This prediction navigation drives the presumed access to increment G+ sub-grid data based on the elementary boundary (that is, depending on the unit transition) of looking of the correspondence that runs into during the prediction navigation.

Now turn to Figure 44, in some embodiments, treatment scheme is since 4401, and wherein all triangles with increment G+ sub-grid are placed in the tabulation that is called " triangle ".Usually the triangle that reads increment G+ sub-grid from non-internal memory (disk) storer [for example, at distributed clients-server implementation (Figure 46, mark storage 4605) in, or in client and server operates in enforcement (Figure 47, mark storage 4705) on the identical physical machine].

Treatment scheme marches to step 4403 and refers to symbol " triangle " so that first triangle in " triangle tabulation " is arranged to variable.

Treatment scheme marches to step 4406, wherein leg-of-mutton Global ID (as being stored in the Triangle data structure of Figure 31 C) is read among the variable GLOBAL_ID.

Treatment scheme marches to decision steps 4408 and determines according to the Global_to_DDL_Linkmap_Tri data structure of Figure 43 A with the DDL ID(that determines " triangle ") whether equal 0 value.Current not being stored in of 0 value representation " triangle " comprises type DDL_Tri(Figure 43 A) the triangle DDL of array of data structures in.Because all values on the Global_to_DDL_Linkmap_Tri array is initialized to 0 value, this sets up.

If determine that in decision steps 4408 value for current " triangle " of Global_to_DDL_Linkmap_Tri array is not 0, this is corresponding to following situation: current " triangle " has been stored in (at index LOCALID place) in the DDL_Tri array, and processes and march to step 4410 and 4412.

In step 4412, the current TRI of increment G+ sub-grid is stored among the VM/PVS that looks the unit for correspondence, above-mentioned correspondence look the unit storage as the LOCALID of the index of the DDL_Tri array of data structures that comprises triangle DDL.

Treatment scheme marches to decision steps 4414 to determine whether remain any untreated triangle in " triangle tabulation ".

If determine not remain untreated triangle in decision steps 4414 in " triangle tabulation ", then treatment scheme marches to step 4416, wherein next increment G+ sub-grid is processed.

On the other hand, if determine the untreated triangle of residue in " triangle tabulation " in decision steps 4414, then treatment scheme marches to step 4418, wherein selects the next triangle in " triangle tabulation " to process.

If determine that in decision steps 4408 value for current TRI of Global_to_DDL_Linkmap_Tri array is 0, then this is corresponding to following situation: current TRI also is not stored among the DDL_Tri, and treatment scheme marches to step 4420 and 4422.

In step 4422, the untapped position of the array of DDL_Tri structure (comprising triangle DDL) is identified and variable OPENINDEX is arranged to the index of this untapped array element.

In subsequent step 4424, with Global_to_DDL_Linkmap_Tri[GLOBAL_ID] value be arranged to OPENINDEX, therefore represent that current TRI will be stored in the OPENINDEX index of triangle DDL.

In subsequent step 4426, the TRI data are stored in the DDL_Tri array of data structures, and will use increment G+ sub-grid data to be added into for index OPENINDEX and the triangle of the VM/PVS data of looking the unit of correspondence and be arranged to refer to the DDL_Tri data, and treatment scheme proceeds to processes 44-1, and it is from the step 4428 of Figure 44 B.

In step 4428, the limit of current " triangle " is stored in " limit tabulation ".

In subsequent step 4430, variable " limit " is arranged to refer to article one limit of " limit tabulation ".

In subsequent step 4432, variable EGLOBAL_ID is arranged to the Global ID on limit by the Edge.global_id field that refers to the limit data structure that comprises increment G+ sub-grid data structure, in Figure 31, specified above-mentioned two kinds of data structures.

The DDL ID(that determines " limit " in follow-up decision steps 4434 determines according to the Global_to_DDL_Linkmap_Edge data structure of Figure 43 A) whether equal 0 value.0 value representation EDGE is current not to be stored among the limit DDL of the array of data structures (Figure 43 A) that comprises type DDL_Edge.

0 value because the upper all values of Global_to_DDL_Linkmap_Edge array all is initialised, this set up.

If what determine the Global_to_DDL_Linkmap_Edge array in decision steps 4434 is not 0 for working as value in front, then this is corresponding to following situation: when the front has been stored in (at index ELOCALID place) in the DDL_Edge array, and treatment scheme marches to step 4438 and 4440.

In step 4440, current " limit " of increment G+ sub-grid is stored in for correspondence and looks among the VM/PVS of storage ELOCALID of unit, and ELOCALID is the index that comprises the DDL_Edge array of data structures of limit DDL.

In follow-up decision steps 4442, determine in " limit tabulation ", whether to remain any untreated limit.

If in decision steps 4442, determine in " limit tabulation ", not remain any untreated limit, then process and be back to step 4414, wherein determine in " triangle tabulation ", whether to remain any untreated triangle.

On the other hand, if determine the untreated limit of residue in " limit tabulation " in decision steps 4442, then treatment scheme marches to step 4443, wherein selects the next limit in " limit tabulation " to process.

If in decision steps 4434, determine Global_to_DDL_Linkmap_Edge for being 0 when value in front, then this is corresponding to following situation: current EDGE also is not stored in the DDL_Edge array, and treatment scheme marches to step 4444 and 4446.

In step 4446, untapped position in the DDL_Edge structure array of (comprising limit DDL) is identified and variable EOPENINDEX is arranged to the index of this untapped array element.

In subsequent step 4448, with Global_to_DDL_Linkmap_Edge[EGLOBAL_ID] value be arranged to EOPENINDEX, the EOPENINDEX index place of the limit DDL that therefore indicates current " limit " is stored in.

In subsequent step 4450, the EDGE data are stored among DDL_Edge array of data structures and the index EOPENINDEX, and the VM/PVS that has added increment VM/PVS data (with the forms of increment Gplussubmesh_edge data) is arranged to refer to corresponding DDL_Edge array.

Subsequently, treatment scheme proceeds to processes 44-3, and it is from the step 4452 of Figure 44 C.

In step 4452, the summit on current " limit " is stored among the VLIST.

In follow-up step 4454, variable " summit " is arranged to refer to first summit of " vertex list ".

In subsequent step 4458, variable V GLOBAL_ID is arranged to the Global ID of VERT by the Vertex.global_id field that refers to the Vertex data structure that comprises increment G+ sub-grid data structure, in Figure 31 C, specified above-mentioned two kinds of data structures.

The DDL ID(that determines " summit " in follow-up decision steps 4460 determines according to the Global_to_DDL_Linkmap_Vertex array of data structures of Figure 43 A) whether equal 0 value.Current not being stored in of 0 value representation " summit " comprises type DDL_Vertex(Figure 43 A) the summit DDL of array of data structures in.Because all values on the Global_to_DDL_Linkmap_Vertex array is both initialized to 0 value, this sets up.

If determine that in decision steps 4460 value for current " summit " of Global_to_DDL_Linkmap_Vertex array is not 0, then this is corresponding to following situation: current " summit " has been stored in (at index VLOCALID place) in the DDL_Vertex array, and treatment scheme marches to step 4462 and 4464.

In step 4464, current " summit " of increment G+ sub-grid be stored in for correspondence look among the VM/PVS of storage VLOCALID of unit, VLOCALID is the index that comprises the DDL_Vertex array of data structures of summit DDL.

In follow-up decision steps 4468, determine in " vertex list ", whether to remain any untreated limit.

Determine not remain in " vertex list " untreated summit in decision steps 4468, then treatment scheme marches to and processes 44-2, and it is back to step 4442 with treatment scheme, wherein determines whether to remain any untreated limit in " limit tabulation ".

On the other hand, if determine the untreated summit of residue in " vertex list " in decision steps 4468, then treatment scheme marches to step 4469, wherein selects the next summit in " vertex list " to process.

If what determine the Global_to_DDL_Linkmap_Edge array in decision steps 4460 is 0 for working as value in front, then this is corresponding to following situation: current VERT also is not stored in the DDL_Vertex array, and treatment scheme marches to step 4470 and 4472.

In step 4472, untapped position in the DDL_Vertex structure array of (comprising summit DDL) is identified and variable V OPENINDEX is arranged to the index of this untapped array element.

In subsequent step 4474, with Global_to_DDL_Linkmap_Vertex[VGLOBAL_ID] value be arranged to VOPENINDEX, the VOPENINDEX index place of the summit DDL that therefore indicates current VERT is stored in.

In subsequent step 4478, the VERT data are stored among DDL_Vert array of data structures and the index VOPENINDEX, and the VERT data that will be added into VM/PVS are arranged to refer to DDL_Vert array corresponding to VOPENINDEX place.

Then, treatment scheme marches to foregoing step 4468.

The method of Figure 44 A to Figure 44 C shows the increment G+ sub-grid data corresponding with triangle, limit and the summit of triangular mesh.Similarly, use to be combined in U.S. Patent No. 6,111, its content of 582(merges among the application by reference) the additional data structure of XIII manage its graphic element such as texture etc.

Figure 45 A, Figure 45 B and Figure 45 C show and support to use the increment VM/PVS(increment G-sub-grid that sends from remote server) the increment VM/PVS of data data storage and the transmission method safeguarded.

Increment G-sub-grid data are corresponding to for specifically looking the become Global ID of the graphic element that recently is blocked of unit transition.At step 4715(or 4725) convert these Global IDs to the DDL index.Can substitute or except the preceding method by usage flag profile contour comes the graphic element that recently is blocked determined, this increment G-sub-grid information that prestores is used for " avoiding " in working time the element that is blocked.

Figure 45 A to Figure 45 C shows to use with specific and looks increment G-sub-grid information exchange that the unit is associated and cross to remove in the corresponding unit transition period specific graphic elements that is blocked that becomes of looking and upgrade the VM/PVS data.

Now turn to Figure 45 A, in some embodiments, treatment scheme is since 4501, and wherein all triangles with increment G-sub-grid are placed in the tabulation that is called " triangle tabulation ".Usually from non-internal memory (disk) storage, read all triangles of increment G-sub-grid.For example, in distributed clients-server implementation (among Figure 46 mark storage 4605), or in client and server operates in enforcement (mark storage 4705 among Figure 47) on the identical physical machine.

In follow-up step 4503, first triangle in " triangle tabulation " is arranged to variable and refers to symbol " triangle ".

In subsequent step 4506, leg-of-mutton Global ID (as being stored in the Triangle data structure of Figure 31 C) is read among the variable GLOBAL_ID.

In follow-up decision steps 4508, determine whether to remove specific " triangle ", still referred to by other VM/PVS arrays (use DDL_TRI[LOCALID] the .num_using field determines, this field is maintained as the quantity of the VM/PVS data structure that equals specific " triangle " in the current DDL_TRI of the referring to array) such as " triangle " in the being stored in DDL_TRI array.When triangle refers to (recently exposing) by the VM/PVS data recently, increase the value for this variable of DDL_TRI, and when corresponding triangle does not refer to (recently being blocked) by the VM/PVS data recently, reduce the value of this variable.

If in decision steps 4508, determine DDL_TRI[LOCALID] value that reduces of .NUM_USING value-1() be not equal to 0, then treatment scheme marches to step 4510.This is corresponding to following situation: though with " triangle " after specific VM/PVS removes, it still appears among other VM/PVS.

In step 4512, the index of " triangle " among the specific VM/PVS is identified and it is read in the variables A 1.

In subsequent step 4513, be that sky removes triangle " triangle " from specific VM/PVS array by the element representation at index AI place with array.

In follow-up decision steps 4514, determine in " triangle tabulation ", whether to have remained any untreated triangle.

If determine not remain untreated triangle in decision steps 4514 in " triangle tabulation ", then treatment scheme marches to step 4516, wherein next increment G-sub-grid is processed.

On the other hand, if determine the untreated triangle of residue in " triangle tabulation " in decision steps 4514, then treatment scheme marches to step 4518, wherein selects the next triangle in " triangle tabulation " to process.

If in decision steps 4508, determine DDL_TRI[LOCAID] value that reduced of .NUM_USING value-1() equal 0, then treatment scheme marches to step 4520.

Step 4520 is corresponding to following situation: removing " triangle " afterwards from specific VM/PVS, it is no longer referred to by any other VM/PVS.

In step 4522, the value of OPENINDEX is arranged to the index for TRI of DDL_TRI, this concrete array element of expression DDL_TRI is not used now.

In subsequent step 4524, with array element GLOBAL_TO_DDL_LINKMAP_TRI[GLOBALID] value be arranged to 0, be illustrated in the DDL_TRI array element at expression specific " triangle ".

In subsequent step 4526, the index of particular VM/PVS " triangle " is identified and it is read variables A I.

In subsequent step 4527, be that sky removes triangle " triangle " from specific VM/PVS array by the element representation at index AI place with array.

Treatment scheme marches to from step 4527 and processes 45-1, and it is from step 45-1(Figure 45 B).In step 4528, all limits of triangle " triangle " all are placed in the tabulation that is called " limit tabulation ".

In subsequent step 4530, variable reference symbol EDGE is arranged on article one limit of " limit tabulation ".

In subsequent step 4532, Global ID's (as being stored in the Edge data structure of Figure 31 C) on limit is read among the variable GLOBAL_ID.

Determine in follow-up decision steps 4534 whether specific " limit " will remove, still referred to by other VM/PVS arrays (use DDL_EDGE[LOCALID] the .num_using field determines, this field is maintained as the quantity of the VM/PVS data structure that equals the specific EDGE in the current DDL_EDGE of the referring to array) such as " limit " in the being stored in DDL_EGGE array.When the limit refers to (recently exposing) by the VM/PVS data recently, increase the value for this variable of DDL_EDGE, and when corresponding limit does not refer to (recently being blocked) by the VM/PVS data recently, reduce the value of this variable.

If in decision steps 4534, determine DDL_EDGE[LOCALID] value that reduces of .NUM_USING-1() be not equal to 0, then treatment scheme marches to step 4538.This is corresponding to following situation: even after specific VM/PVS has removed EDGE, it still is present among other VM/PVS.

In step 4538, the index on " limit " among particular VM/PVS is identified and it is read among the variables A I.

In subsequent step 4539, be that sky removes limit " limit " from specific VM/PVS array by the element representation at index AI place with array.

In follow-up decision steps 4542, determine in " limit tabulation ", whether to remain any untreated limit.

If in decision steps 4542, determine in " limit tabulation ", not remain untreated limit, then process and be back to step 4514.

On the other hand, if determine the untreated limit of residue in " limit tabulation " in decision steps 4542, then treatment scheme marches to step 4543, wherein selects the next limit in " limit tabulation " to process.

If in decision steps 4534, determine DDL_EDGE[LOCALID] value that reduced of value-1(of .NUM_USING) equal 0, then treatment scheme marches to step 4544.

Step 4544 is corresponding to following situation: with " limit " after specific VM/PVS removes, it is no longer quoted by any other VM/PVS.

In step 4546 value of OPENINDEX is arranged to DDL_EDGE for the index of EDGE, this concrete array element of expression DDL_EDGE is not used now.

In subsequent step 4548 with array element GLOBAL_TO_DDL_LINKMAP_EDGE[GLOBALID] value be arranged to 0, be illustrated in the DDL_EDGE array element on expression specific " limit ".

In follow-up step 4550, the index on " limit " among the specific VM/PVS is identified and it is read among the variables A I.

In subsequent step 4551, be that sky removes limit " limit " from specific VM/PVS array by the element representation at index AI place with array.

The treatment scheme row marches to from step 4551 and processes 45-3, and it is from step 4552.In step 4552, all summits of triangle " triangle " are prevented in the tabulation that is called " triangle tabulation ".

In subsequent step 4554, the first side in " triangle tabulation " is arranged to variable reference symbol VERT.

In subsequent step 4558, Global ID's (as being stored in the Vertex data structure of Figure 31 C) on summit is read variable GLOBAL_ID.

In follow-up decision steps 4560, determine specificly whether will remove specific VERT, still referred to by other VM/PVS arrays (use DDL_VERTEX[LOCALID] the .num_using field determines, this field is maintained as the quantity of the VM/PVS data structure that equals the specific VERTEX in the current DDL_VERTEX of the referring to array) such as the VERT in the being stored in DDL_VERTEX array.When the summit refers to (recently exposing) by the VM/PVS data recently, increase the value for this variable of DDL_EDGE, and when corresponding summit does not refer to (recently being blocked) by the VM/PVS data recently, reduce the value of this variable.

If in decision steps 4560, determine DDL_VERTEX[LOCALID] value that reduced of .NUM_USING value-1() be not equal to 0, then treatment scheme marches to step 4562.This is corresponding to following situation: even after specific VM/PVS has removed VERT, it still appears among other VM/PVS.

In step 4562, the index of the VERT among the specific VM/PVS is identified and it is read variables A I.

In subsequent step 4564, be that sky removes summit VERT from specific VM/PVS array by the element representation at index AI place with array.

In follow-up decision steps 4568, determine in " triangle tabulation ", whether to have remained any untreated summit.

If in decision steps 4568, determine in " triangle tabulation ", not remain untreated summit, then process and be back to step 4542.

On the other hand, if determine the untreated summit of residue in " triangle tabulation " in decision steps 4568, then treatment scheme marches to step 4596, wherein selects the next summit in " triangle tabulation " to process.

If in decision steps 4560, determine DDL_VERTEX[LOCALID] value that reduced of value-1(of .NUM_USING) equal 0, then treatment scheme marches to step 4570.

Step 4570 is corresponding with following situation: after specific VM/PVS has removed VERT, it is no longer quoted by any other VM/PVS.

In step 4572, the value of OPNEINDEX is arranged to the index for VERT of DDL_VERTEX, this concrete array element of expression DDL_VERTEX is not used now.

In subsequent step 4574, with array element GLOBAL_TO_DDL_LINKMAP_VERTEX[GOBALID] be arranged to 0, be illustrated in the DDL_VERTEX array element not at the specific VERT of expression.

In subsequent step 4576, the index of the VERT among the specific VM/PVS is identified and it is read among the variables A I.

In subsequent step 4578, be that sky removes summit VERT from specific VM/PVS array by the element representation at index AI place with array.

After step 4578, treatment scheme marches to the decision steps 4568 of having described.

Figure 45 shows the increment G-sub-grid data corresponding with triangle, limit and the summit of triangular mesh to scheming C.Similarly, use to be combined in U.S. Patent No. 6,111,582(merges to it among the application by reference) Table X III in additional data structure manage other graphic elements such as texture etc.

Figure 46 is distributed clients-server implementation, and the external memory increment G sub-grid data of server access storage wherein change Global ID on DDL ID and maintenance server and the client DDL, VM/PVS data.

Figure 46 shows the figure that carries out the embodiment of the client process on independent client device in the server process on the server apparatus and execution.Figure 46 shows the data that send to client device from server apparatus.In some embodiments, this data communication device is crossed local Mesh connection or distributed grid and is connected and transmit such as the Internet.

Server apparatus is represented by the dark-coloured rectangle that is labeled as 4601.Client device is represented by the dark-coloured rectangle that is labeled as 4602.

External memory or disk storage are arranged in server apparatus and are marked as 4605.This memory device, stores comprises the increment G sub-grid data (increment G+ sub-grid data and increment G-sub-grid data) of the increment VM/PVS data corresponding with specifically looking the unit transition.In other embodiment, mark profile contour information is stored and the traversal method that has been used for the simplification of Figure 36 and Figure 37 generates increment G+ or increment G-sub-grid data according to the unified observability figure of correspondence.

When the supposition navigation algorithm (being carried out by server) by current view point position (sending to server from client's section) guiding runs into when looking the unit transition, as as shown in the step 4610, from external memory memory access corresponding increment G+ sub-grid data (corresponding to the graphic element that recently exposes) and increment G-sub-grid data (corresponding to the graphic element that recently is blocked).Additionally and/or alternately, increment G+ and/increment G-sub-grid data can be during looking ahead usage flag profile contour information (with Figure 33 as precomputation with its generation and be stored as outer deposit data) generate.In some embodiments, with the simplification of Figure 37 the profile information processed according to mark of trellis traversal generate increment G+ and increment G-sub-grid data.

With the simplification of Figure 37 traversal method be applied to given unified observability figure (comprise the profile contour of looking unit such as A and B for two connections/block border), construct increment G+ sub-grid according to some embodiments with the leg-of-mutton set of seed, and construct increment G-sub-grid with the leg-of-mutton supplementary set of seed.

For example, consider to comprise for the profile contour of looking unit A and B of two connections and the concrete unified observability figure that blocks the border polyline of correspondence.How following table can generate or walk around increment G+ and increment G-sub-grid with the mark contour of the selection of depending on the set/tabulation of seed triangle if showing.

Table X IV

Come according to generating increment G+ and increment G-sub-grid data for looking unit A with the unified observability figure that looks unit B with the method for Figure 36 and Figure 37

Therefore, because increment G+(and increment G-) the sub-grid data can generate according to the mark profile contour of looking the unit transition for correspondence in working time, so do not need to store increment G+(and increment G-) the sub-grid data.Because compare in many cases increment G sub-grid data, mark contour information can have less carrying cost far away, so the method reduces the overall storage demand fully.This especially sets up in following situation: the profile contour selected is looked unit transition recycling for some, and it is set up when high partly effectively dynamically blocking (EDO), corresponding to the high district of the visible continuity of the local dynamic station of model.This also sets up in following situation: the leg-of-mutton quantity of recently exposing or recently being blocked in the increment G grouping very large (with respect to the complexity of blocking the border that limits corresponding increment district).

According to some embodiments, in order to generate for the increment G+ that looks the unit transition or increment G-sub-grid information in working time, with all polygonal mesh datas (looking as seen from the specific unit of looking) and mark profile contour information (looking the unit transition corresponding to specific) but be presented on quick access (internal memory) storer.

In order to ensure before increment G sub-grid generate to be processed from external memory this information of prestrain, server process directly access is looked more greatly unit look visible polygon set or polygon segmentation from what comprise that will generate increment G sub-grid data specific look the unit.These data are with respect to for the direct PVS that looks the unit, but minimize storage demand with storage for relative little these the direct PVS data of looking greatly the unit of quantity by precomputation, according to the involved PVS that looks the unit that looks greatly the unit and during looking ahead, uses the trellis traversal processing simplified and mark profile contour information to be paired in larger quantity next life.Alternative, can before be configured to the PVS that large father looks unit itself for the adjacent existing VM/PVS that looks greatly unit (being that the father that other sons are looked the unit looks the unit) structure itself for the father to father's the enhancing G sub-grid of looking the unit transition is data based by access.

In addition, server process is also accessed the mark profile contour information for the particular sub-unit that will generate increment G sub-grid information.

When being presented on internal memory for polygonal mesh information and the mark profile contour information of constructing increment G sub-grid data, the external memory access does not interrupt being undertaken by server process the generation of increment G sub-grid data.

The use that the father and son looks cell data be utilize exist and hierarchy unit between the example of simple observability hierarchical structure: look all polygon/segmentations of looking the unit from son and also look the unit and look as seen from comprising the father.Look in this application of unit the father, comprise that it is very important in fact being positioned at all polygon/segmentations that the father looks the unit.Finally, usually the terminal leaf is looked the unit and construct and it is comprised seldom or do not comprise geometric figure, so that their VM/PVS exclusively " looks from looking the unit " and less " is comprised in and looks the unit ".

Because that can determine in advance any grid looks the surface subdivision level of looking the unit from any, so the method for mark profile contour easily is suitable for the face of refinement.

The other details of this Gradation processing has been proposed in conjunction with Figure 52.

In subsequent step 4615, determine the DDL index of the element that each is accessed with approximate Global_to_DDL_Linkmap.

Whether the data of determining recently access in decision steps 4620 have been presented among the corresponding DDL (about the details of geometric graphic element, referring to the step 4408,4434 and 4460 of Figure 44).

If determine that in decision steps 4620 data of accessing have been presented among the corresponding DDL, then treatment scheme directly marches to step 4625.

In step 4625, the DDL index that is pre-existing in of usage data copies server copy and the client that the data of accessing are written to approximate VM/PVS data among both.

On the other hand, if determine that in decision steps 4620 data of accessing also are not presented among the corresponding DDL, then treatment scheme marches to step 4630.

In step 4630, empty position to the server copy of corresponding DDL is identified, and in step 4635, the data of accessing are write the server copy of DDL and client copy this position on both (for example, referring to the step 4420 of Figure 44 to step 4426).

Specific graphic elements among the DDL (for example, triangle or texture) often is presented among in the VM/PVS data that are stored on the client device more than one at any one time.This is that client is often stored and approximating VM/PVS data acquisition corresponding to unit of looking because use supposition navigation forecasting method.

Because server process at first checks with the increment G sub-grid data determining to be accessed and had appeared in the past among the corresponding DDL, so avoided redundant transmission data being sent to client.This has reduced the required bandwidth that is connected between client and the server fully.

Figure 47 shows the block diagram/flow diagram that server process is positioned on the identical equipment with client process and uses the layout of single part of DDL copy.

Figure 47 shows the figure that carries out the server process on the computing equipment identical with the computing equipment of client process.This configuration deposit data except increment G+ sub-grid and increment G-sub-grid is identical with the configuration of Figure 46 beyond being stored on the equipment identical with the equipment of client rendering system.

The needs to the client and server copy of DDL and VM/PVS data have been eliminated in this configuration.4710 to 4725 processing is identical except following: the portion copy that only needs to write corresponding data.

In this case, the major function of DDL is to reduce the carrying cost of VM/PVS and simplify Global ID to the conversion of local array (DDL) index.

Figure 48 A shows to send and reduces the detailed increment visibility information of level in case current view point will soon penetrate also the block diagram/flow diagram of the server process of looking elementary boundary that does not read/send the increment visibility information.

The block diagram of Figure 48 A and process flow diagram show the additional detail of the client-server processing of transmitting the increment visibility information.No matter client and server is to be implemented in the different physical equipments that connect by network connection (as in Figure 46), or client and server is to be implemented in (as in Figure 47) in the identical physical equipment, can implement these steps.

In Figure 48 A, block diagram is divided into different processing with the client and server function, and does not consider whether these processing are carried out on different physical equipments.

Be labeled as 4802 frame and comprise client process, be labeled as 4855 frame and comprise server process.

In the first step, client process obtains current view point and current view point is stored in the Data Position 4805.Can come the data that provided the current view point position are made amendment by the input equipment that comprises mouse, control lever, game control pad or revise other input equipments of current view point position that is connected to client process.

Client process will offer server process to 4805 entrance of data position, its with data copy in Data Position 4810.This copy can occur in local storage or run through grid.

In step 4815, server process is predicted the Future Positions of viewpoint with the current view point data.This prediction can be adopted dead reckoning and collision constraint, velocity and acceleration restriction and according to the additive method of the following viewpoint position of current view point position prediction, for example is generally used for determining the method for navigation of prior art of non-player gladiator or artificial intelligence agency's navigation in counting game or other simulators.(referring to " Tactical Pathfinding Using A* ", Van Der Sterren, William in Game Programming Gems3, Charles River Media2002, its full content merges among the application by reference).

In step 4820, that uses that the following viewpoint position of predicting determines to predict looks the unit transition, is to looking passing of elementary boundary depending on the unit transition.This step adopts from the data of step 4815 with for the known arrangement of looking the unit of model.Can read by the server process in the step subsequently the increment observability data (being also referred to as visibility events information) of looking elementary boundary (that is, at every turn looking the unit transition) correspondence of passing with each.

In step 4825, the arrival for current view point was determined with the time of looking elementary boundary corresponding to unit transition of looking minimum of predicting of processing.This time is current view point position and speed, the acceleration of maximum possible and the function of speed and local collision and other navigation constraints.This value is stored in the variable " time of break-through ".

In step 4830, the maximum time of reading and transmit the increment visibility information corresponding with looking the unit transition future of processing is determined.This value is based on corresponding increment observability grouping, determines to the available transmission bandwidth of client and other possible factors.This value is stored in the variable " transmission time ".

Determine that in follow-up decision steps 4835 whether variable " transmission time " is greater than variable " time of break-through ".If determine variable " time of break-through " greater than " transmission time " in decision steps 4835, then treatment scheme marches to step 4840.

In step 4840, (for example from external memory, read the increment observability data of low level of detail (LOD) version, increment G+ sub-grid data) and make it can be used for client process, it comes with to look the unit corresponding and be stored in the observability figure that is labeled as in 4850 the Data Position/PVS data and upgrade with these data.In embodiment, this relatively the increment observability data of low level of detail comprise the texture resolution of geometric figure details with level relatively, lower level or additive method minimizing the data of details.Other embodiment comprises that the information that is applied to this lower level of detail is to shield the wave filter of low level of detail feature.For example, rule model can comprise by surpassing 100 ten thousand polygonal meshs that mesh triangles shape forms.In some embodiments, the low level of detail version of regular image can only comprise 10% mesh triangles shape.For the existing method that generates the model on the different level of detail, referring to Leubke, David; Reddy, Martin et.al, " Level of Detail for3D Graphics " Morgan Kauffman2003, ISBN:1-55860-838-9, its full content merges among the application by reference.

If determine that in decision steps 4835 variable " transmission time " is not more than variable " time of break-through ", then treatment scheme marches to step 4845.

In step 4845, the increment observability data that read higher level of detail version from external memory (for example, increment G+ sub-grid data) and make it can be used for client process, it uses this data pair and to look the unit corresponding and be stored in the observability figure that is labeled as in 4850 the Data Position/PVS data and upgrade.

Step 4815 is guaranteed to 4840: only when upgrading all increment visibility information required for the PVS that looks the unit received by client process before current view point when probably entering this and looking the unit, just adopt geometry information and/or the texture information of the relatively low level of detail of increment observability in dividing into groups.

Under situation, the use of lower level of detail information reduced the increment visibility information read and transferred to the required time of client process, thus reduced increment be configured to look the unit all required increment visibility information of PVS be read and transmit before viewpoint enter the possibility that this looks the unit.

In addition, in this case use to this low level of detail information is unlikely detected by the user.This is that corresponding to the face that recently exposes, it has exposed the very of short duration time in this case because of the increment visibility information (for example, increment G+ sub-grid information) that provides; And below be known: the human visual system has low visual sensitivity for the mask that has exposed blink.

Figure 49 shows as exposing the time and exposing contrast sensitivity's's (measurement of visual sensitivity) of the function in district figure, in conjunction with this figure it has been carried out detailed discussion.

Therefore, this method is following methods: initially come current view point is upgraded at the PVS that looks the unit that can provide corresponding higher level of detail probably to penetrate in the past with the increment visibility information of low level of detail relatively.In embodiment, send relatively higher level of detail increment visibility information in the time subsequently, such as exposing the time that increment G+ sub-grid face increases, expose the time corresponding to following: the time dependence spatial sensitivity meeting of user's vision system is so that the user understands geometry and/or the texture information of higher level of detail.In some embodiments, the time of the visibility information of the higher level of detail of transmission is subjected to the restriction of the ability of human perception object.

The compare information of high level of detail, the graphical information of low level of detail is passed to client unit more rapidly.When the information of passing through to transmit low level of detail when client is threat during the of short duration time period that delays to reach of graphical information, this method has prevented delaying to reach of graphical information fully.Delaying to reach of graphical information can be thought the not failure of expectation of perception ground based on the pre-fetching system of observability, this is because the graphic element (or their unexpected reproductions afterwards) that misses is easy to that the observed person perceives and be that vision ground is disgusting.

By initially during these time periods, transmitting the graphical information of low level of detail and sending subsequently the additional patterns information that can be used in the level of detail that increases initial graphics information, the method can prevent delaying to reach of graphical information, and still can realize mis-behave harmless on the consciousness when visual sensitivity VS shown in Figure 49 exposes time curve when the replacement timetable mates or surpasses.

If the replacement timetable exposes time curve near visual sensitivity VS shown in Figure 49, then the user can experience the mis-behave of appropriateness in the perception.This mis-behave can be experienced replace low level of detail graphical information with higher level of detail graphical information during is discrete replacement.Along with the replacement timetable exposes time curve near the visual sensitivity VS of Figure 49, the more difficult perception for the user that can become of this replacement time.Under any circumstance, replacement event surpasses the delaying to reach of graphical information (and the follow-up unexpected reproduction that can be caused by delaying to reach of graphical information).

Thereby the method has produced the mis-behave of appropriateness in harmless in the perception or the perception during visual momently in the situation of the low visible continuity of space-time.This visual momently low visible continuity of space-time may by the observability of model often with the momently high viewpoint speed in the district (near for example, the district the door from the indoor moving to the open air) that changes at a high speed or partly low visible correlativity cause.The method also is useful processing following situation: available transmission bandwidth reduces momently, or server performance reduces momently.

Be under the situation of high viewpoint speed in a kind of important situation, the time of exposing of increment G+ sub-grid face may not can increase to so that the user can differentiate the duration of the face that recently exposes fully.This is that the time that the face that recently exposes remains in during the viewport can be of short duration because under high viewpoint speed.

Because the method for Figure 46 and Figure 47 adopts the speculative prefetching based on the increment visibility information of the following viewpoint position of looking ahead, so high viewpoint speed tends to produce mis-behave, this is because for high viewpoint speed, and the increment visibility information amount that must read and transmit can surpass the reading capability of server or available transmission bandwidth.

This method of Figure 48 A is by naturally so that the mis-behave of this method relatively is difficult to the mode of perception comes the performance of this method and human visual system's performance are mated to minimize these mis-behaves.

Because this method adopts looking ahead and high-speed cache based on the visibility events data of navigation prediction, so usually visibility events grouping is sent to client unit from server unit, this is to have passed the viewpoint itself of looking unit transition boundary rather than user and passed and look the unit transition boundary because be actually navigation prediction algorithm agency as the viewpoint of predictive user.

Because this speculative prefetching, in fact section blink of the visible correlativity of many low spatials of measuring according to user's viewpoint can not produce delaying to reach of visibility events grouping, this be because these identical meeting in groups since the speculative prefetching (being also referred to as looking ahead based on observability) of being eager to navigate be sent out and high-speed cache.

Only when the visibility events grouping that postpones is in fact predicted, determined in 4830 such as the step 4820 of Figure 48 A, this method sends the information of low level of detail to prevent that postponing grouping arrives.

Figure 49 shows the figure of the impact of exposing the time differentiates spatial detail on the mankind ability.

Figure 49 is the data (Luntinen that has the human vision performance study to reproduce, O., et.al. (1995) .Modeling the Increase of Contrast Sensitivity with Gating Areaand Exposure time.Vision Res.Vol.35, No.16, pp.2339-2346.ElsevierScience Ltd.).

This studies show that and exposed the time for the remarkable result of contrast sensitivity, especially for until the high spatial frequency of 1000ms.

Figure 49 shows 4 width of cloth figure (A to C), and wherein contrast sensitivity is drawn into the function that exposes the time from 0 second to 10 seconds.In every width of cloth figure, relation is to draw for the different area that always exposes of observed face (comb mesh pattern).Every width of cloth figure represents the relation of one of four spatial frequencys for represented.This studies show that and few exposes the time and the little area that exposes has reduced contrast sensitivity.Note, to increase along with the increase of spatial frequency be because measured low frequency is lower than " peak value " spatial frequency about the spatial contrast sensitivity curve under the experiment condition that adopts to contrast sensitivity in this case.Be apparent that according to these data, the ability that the human visual system differentiates high spatial frequency data is even surpasses the function that exposes the time of 1000ms.This meeting so that low level of detail increment visibility information purposes by initially with and the increment visibility information of the relatively high level of detail that exposes the time of the increment G+ sub-grid data of the follow-up correspondence that is used as increase substitute above-mentioned low level of detail increment visibility information.

Whether the method for having few geometric figure/texture that exposes the time and read/transmit the increment observability data of low level of detail of Figure 48 A can be utilized the sensitivity VS that is provided by Figure 49 to expose relation data and should and when be read/transmit in the time subsequently with the increment observability data of determining relatively high level of detail.

Figure 50 A is the block diagram/flow diagram of the server process of the increment visibility information that is high enough to cause the level of detail that has reduced in the situation of the visual sensitivity that has reduced for mobile geometric figure and/or texture of the image-space velocity at the recently visible geometric figure corresponding with the increment visibility information and/or texture.

The block diagram/flow diagram of Figure 50 A shows the additional detail of the client-server processing of transmitting the increment visibility information.No matter client and server is implemented in the different physical equipments that connect by Mesh connection (as in Figure 46), or the client and server processing is implemented in (as in Figure 47) in the identical physical equipment, can both implement these steps.

Among Figure 50 A, block diagram is separated into different processing with client process and server process, and does not consider whether these processing are carried out on different physical equipments.

Be labeled as 5002 frame and comprise client process, be labeled as 5055 frame and comprise server process.

In first step, client process obtains current view point and current view point is stored in the Data Position 5005.Can come the data that provided the current view point position are made amendment by the input equipment that comprises mouse, control lever, game control pad or revise other input equipments of current view point position that is connected to client process.

Client process offers server process with the entrance of Data Position 5005, its with data copy in Data Position 5010.This copy can occur in local storage or run through grid.

In step 5015, server process is predicted the Future Positions of viewpoint with the current view point data.This prediction can be adopted dead reckoning and collision constraint, velocity and acceleration restriction and according to the additive method of the following viewpoint position of current view point position prediction.

In step 5020, that uses that the following viewpoint position predict determines to predict looks the unit transition, as look elementary boundary pass look the unit transition.This step adopts from the data of step 5015 with for the known cell layout of looking of model.Can read with each elementary boundary of looking that passes by server process in step subsequently be that each looks increment observability data corresponding to unit transition.

In step 5025, to determining with the image-space velocity of geometric element corresponding to increment visibility information (for example, for specific increment G+ sub-grid data of looking the unit transition), and use variable " element _ speed " storage the above results.

In subsequent step 5030, the retina speed of estimated element is determined as the function of pattern space speed, and used variable " estimation _ retina _ speed " storage the above results.Can adopt known being used for that the transport function that the retina speed of the function of the image-space velocity of this element of conduct of element and other factors (for example, the predictability of image-steric acceleration, motion and other possible factors) is predicted is estimated retina speed.

Determine that in follow-up decision steps 5035 whether variable " estimation _ retina _ speed " is greater than predetermined variable " value _ R ".If determine that in decision steps 5035 " estimation _ retina _ speed " greater than " value _ R ", then treatment scheme marches to step 5840.

The increment observability data that read low level of detail (LOD) version from external memory in step 5840 (for example, increment G+ sub-grid data) and make it can be used for client process, it uses this data pair and to look the unit corresponding and be stored in the observability figure that is labeled as in 5850 the Data Position/PVS data and upgrade.

If determine that in decision steps 5835 variable " estimation _ retina _ speed " is not more than variable " value _ R ", then treatment scheme marches to step 5845.

In step 5845, the increment observability data that read higher level of detail version from external memory (for example, increment G+ sub-grid data) and make it can be used for client process, it uses this data pair and to look the unit corresponding and be stored in the observability figure that is labeled as in 5850 the Data Position/PVS data and upgrade.

Step 4815 to step 4840 only guarantee when recently visible geometric figure and/or texture have produce reduce the observer for the image of the retina speed of the spatial vision sensitivity of visible element recently-space velocity the time, just adopt the increment observability divide into groups in relatively low level of detail geometry information and/or texture information.

High viewpoint speed can produce the high image-space velocity of geometric graphic element.As previously discussed, in the situation that high viewpoint speed, current view point is probably advanced into this for all required increment visibility information of the PVS that looks the unit by the client process reception in renewal and looks the unit.

In this case the use of low level of detail information has been reduced to client process and read and transmit the required time of increment visibility information, enter the possibility that this looks the unit thereby reduced viewpoint before can being read/transmitting in all required increment visibility information of PVS that Incremental Construction is used for looking the unit.

In addition, detected by the user the use of low level of detail information is unlikely in this case.This is because of geometric graphic element and/or the texel of the increment visibility information (for example, increment G+ sub-grid information) that provides corresponding to the retina speed of the height prediction of the hi-vision-space velocity of the ability with the resolution spatial detail shown in Figure 51 that reduces the observer and correspondence.

Figure 50 B is the alternate embodiments of the method shown in Figure 50 A.In Figure 50 B, process the level of detail that determines when the graphical information that reduction is sending with different tests.In this embodiment, employed test was not based on the retina speed of estimating, sent the required bandwidth of visibility events data and the comparison of available bandwidth and be based on.

In Figure 50 A, block diagram is separated into different processing with client functionality and server capability, and does not consider whether these processing are carried out on different physical equipments.

Be labeled as 5052 frame and comprise client process, be labeled as 5099 frame and comprise server process.

In first step, client process obtains current view point and current view point is stored in Data Position 5055.Can come the data that provided the current view point position are made amendment by the input equipment that comprises mouse, control lever, game control pad or revise other input equipments of current view point position that is connected to client process.

Client process offers server process with the entrance of Data Position 5055, its with data copy in Data Position 5060.This copy can occur in this external memory or run through network.

In step 5065, server process is predicted the Future Positions of viewpoint with the current view point data.This prediction can be adopted dead reckoning and collision constraint, velocity and acceleration restriction and according to the additive method of the following viewpoint position of current view point position prediction.

In step 5070, that uses that the following viewpoint position predict determines to predict looks the unit transition, as look elementary boundary pass look the unit transition.This step adopts from the data of step 5065 with for the known cell layout of looking of model.Can read with each elementary boundary of looking that passes by server process in step subsequently be that each looks increment observability data corresponding to unit transition.

In step 5075, transmission is determined based on the required lasting bandwidth of visibility events data of the following viewpoint position of determining to predict in step 5070.This can penetrate the speed of looking elementary boundary together with the navigation prediction algorithm according to the known packet size of the visibility events data corresponding with looking the unit transition and determine.

In step 5080, determine available bandwidth.

In follow-up decision steps 5085, bandwidth demand and the available bandwidth that continues compared.

Surpass available bandwidth if determine the required bandwidth that continues in decision steps 5085, then treatment scheme marches to step 5090.

In step 5090, server sends the increment observability data (visibility events data) that are in low fully level of detail.This required bandwidth that can reduce to continue postpones grouping arrival with alleviating.

If determine that in decision steps 5085 the required bandwidth that continues does not exceed available bandwidth, then treatment scheme marches to step 5095.

In step 5095, send the visibility events data with relatively high or normal level of detail.

Be used for the required bandwidth that continues that visibility events stream processes often with being inversely proportional to of visual temporary transient visible continuity.The lasting condition of low temporary transient visible continuity can be produced by relatively high user's viewpoint speed, relatively high user's viewpoint speed is easy to produce the relatively high retinal patterns speed for visual graphic element, as discussing in conjunction with Figure 50 A.

As in the situation of the momently low temporary transient visible continuity of discussing in conjunction with Figure 48 A and Figure 49, when the visible continuity that continues increases, can send additional graphical information with the level of detail of the graphic element that increases previous transmission with harmless, progressive geometric figure and texture compression method.

Again, can be by the scheme of looking ahead based on observability of the navigation prediction algorithm driving of the data cache that will continue on client unit because this method has adopted, so level of detail that can be low initially sends graphical information, then its become the user visible before with its replacement.

Figure 51 shows the figure as the relative spatial vision sensitivity of retina speed, in conjunction with this figure detailed discussion has been carried out in this relative spatial vision sensitivity.

Therefore, this method is to use the relatively method of the increment visibility information of low level of detail corresponding with the recently visible element with high prediction retina speed.

Because the method for Figure 46 and Figure 47 adopts the speculative prefetching based on the following viewpoint position of predicting of increment visibility information, so high viewpoint speed tends to produce hydraulic performance decline, this is because for high viewpoint speed, and the increment visibility information amount that must read and transmit can surpass the reading capability of server or available transmission bandwidth.

This method of Figure 50 A is by being that dynamic vision sensitivity mates to minimize these hydraulic performance declines with the performance of this method and another performance limitations of human visual system.Dynamic vision sensitivity has reflected the spatial sensitivity that has reduced for the element that has relatively high retina speed.Thereby the method is easy to produce the hydraulic performance decline of appropriateness.

Figure 48 shows with the navigation restriction and avoids the visibility events grouping that postpones and the method for recovering client VM/PVS or increment G data.

The method of Figure 48 A, Figure 50 A and Figure 50 B was used for hanging down the packetization delay that the level of detail grouping helps prevent institute's perception by sending during the time period of low temporary transient observability correlativity.Usually, the specified method of this paper adopts the looking ahead based on navigation of visibility events grouping (for example, increment G grouping).Prevent that the alternative method that packetization delay arrives from being to adjust observer's navigation speed.By the viewpoint speed that slows down, can increase temporary transient visible continuity.

During the application of roaming type, some factors can limit navigation speed naturally.Such as muscular fatigue, engine is overheated or the factors such as insufficient reception that are used for the remote signal of control navigation are easy to limit during the high-speed navigation or tight speed after it naturally.These high-speed navigation time periods are corresponding to the time period of low temporary transient visible correlativity, the viewpoint navigation may surpass the limit of the visibility events grouping of looking ahead during this time period, from causing the delay transmission of grouping or exhausting of client VM/PVS data cache.In the method, can seamlessly regulate the client cache of viewpoint speed to be easy to the preventing visibility events packetization delay and to recover VM/PVS or increment G data with these factors.

Figure 48 B shows and applies the navigation restriction to prevent the exemplary process diagram of the method that delays to reach that visibility events is divided into groups.

Now turn to Figure 48 B, in some embodiments, treatment scheme from decision steps 4850, wherein determine the visibility events data whether can accessed by client unit and show before arrive.

If determine that in decision steps 4860 visibility events information can not delay to reach, then end process.

On the other hand, can be scheduled with arrival after being accessed by client unit/showing if determine the visibility events data, then treatment scheme marches to step 4865, wherein reduces viewpoint speed.On client unit, allow to viewpoint speed reduce reduced the visibility events information corresponding with visible graphical information recently and be transferred to the necessary speed of client unit, arrive thereby be easy to prevent postpone grouping.

Treatment scheme marches to optional step 4870, wherein so that the indicator of the speed that reduces shown by client unit.This indicator can comprise the power that physics is tired, engine is tired, not enough or navigation hinders as the indicator of the road conditions of barrier or bad luck etc.To the demonstration of the speed state that reduced so that speed for client user's more nature and measurable that seems.

In the embodiment that substitutes, the method for the exemplary process diagram of Figure 48 B can be undertaken by server unit or client unit or the said two units that is operated in together.

Figure 51 shows the figure as the relative spatial vision sensitivity of the function of retina speed.

Can reappear according to the data that provide with Publication about Document the figure of Figure 51: Eckert, M.P.and Buchsbaum, G. (1993) .The Significance of Eye Movements and Image Acceleration for Coding Television Image Sequences.In Digital Images and Human Vision, Watson, A.B. (Ed.), MIT Press, chapter8, pp.89-98.

The resolution that these data show the human visual system has the obvious decline of ability of details of the element of high retina speed.

Dynamic vision sensitivity is that human visual system's the space-time limit is (referring to Brown, B. (1972b) Dynamic Visual Acuity, Eye Movements and Peripheral Acuity for moving targets.Vision Res.Vol.12, pp.305-321.Pergamon Press1972.), it has expressed the sensitivity for Moving Objects of vision system.The graphic element of the light stream on the experience retina receiver produces amphiblestroid of short duration stimulation and may not can satisfy and differentiate the moving picture element higher position reason element of required integral time fully.So vision system is differentiated high retinal patterns speed badly.

The retinal patterns speed of the graphic element that eye motion often reduces to follow the tracks of by the moving eye strategy of chasing and watch attentively.No matter for having high image-space velocity, high image-steric acceleration or the object of uncertain image-spatial movement, it is not the fully accurate and retinal images motion that produces the distinguishable rate that reduces movement elements that moving eye is followed the tracks of.Effectiveness that moving eye chases is described with the retinal images velocity correlation be the relation (such as Lisberger et al. (1987) Annual Review of Neuroscience1987,10:97-129) of the function of image-steric acceleration and predictability be known and be used for based on image-space velocity and/or the acceleration of tracking element estimate retina speed.

Figure 52 shows process flow diagram/block diagram that following client and server is processed: server access look for the father unit pre-stored increment G sub-grid and be configured to the observability figure that the father looks the unit/PVS with this information, then use above-mentioned observability figure/PVS to be configured to the increment G sub-grid data that son is looked the unit transition together with the profile contour information of the pre-stored of looking the unit transition for son.

Compare storage and look the PVS of unit for each, use the increment G+ sub-grid of in this manual description and the increment PVS storage scheme of increment G-sub-grid data to have lower fully carrying cost.Yet, control heuristics even use based on effective static barriers (ESO) and the precision of effectively dynamically blocking (EDO), for having many large databases of looking the unit, for increment PVS data storage cost for often remaining very large.

A kind of method of size that reduces the increment PVS database of precomputation is only to store for relatively large " father " to look the increment G sub-grid data of unit and come to generate in working time the increment G sub-grid data of looking the unit for son with the mark profile contour information of pre-stored.

Introduce this scheme and provide additional details at this in conjunction with the discussion of Figure 46 and Figure 47.

Now turn to Figure 52, Figure 52 shows the figure that carries out the client process on independent client device in the server process on the server apparatus and execution.Figure 52 and Figure 46 are closely similar except the following stated: although in the method for Figure 46, can directly read all increment G sub-grids from external memory.But in the method for Figure 52, directly from storer, read the increment sub-grid data of only looking the unit for the father.Be configured to the VM/PVS that the current parent looks the unit with this father's increment G sub-grid data together with previous father VM/PVS.Come in the increment G sub-grid data of generation working time for subelement together with the unit transition of mark profile information look to(for) son with this new father VM/PVS.

Compare direct storage and look the increment G sub-grid data of unit transition for every second son, the mark profile contour information of the data structure of use Figure 31 and relevant algorithm (Figure 32 to Figure 37) can have less carrying cost.In addition, the simplification of Figure 37 3D trellis traversal algorithm guarantee to construct rapidly in working time the corresponding increment G sub-grid data for subelement.Use the method, can realize the optimum balance between the assessing the cost of carrying cost and working time.

Now turn to Figure 52, server is represented by the dark-coloured rectangle that is labeled as 5201.Client device is represented by the dark-coloured rectangle that is labeled as 5202.

External memory storage in the server apparatus or magnetic disk memory are marked as 5205 and 5207.The increment G sub-grid data (increment G+ sub-grid data and increment G-sub-grid data) that storer 5205 storages are looked the unit transition for specific father.

In step 5210, look unit VM with this data based previous father and construct father's observability figure (VM).

Be labeled as the mark profile contour information of looking the unit transition that the storage of 5207 external memory storer or magnetic disk memory is looked the unit for son.

In step 5214 with these data come with the father look the VM of unit and Figure 37 simplification 3D trellis traversal and associative list XIV be configured to increment G+ sub-grid and the increment G-sub-grid data that son is looked the unit transition.

Figure 36 is connected the 3D polygonal mesh traversal method of simplification is applied to the unified observability figure that provides (comprise for two connections look for example profile contour of A and B/block border, unit) with Figure 37, can construct increment G+ sub-grid with the leg-of-mutton set of seed, and construct increment G-sub-grid with leg-of-mutton this Complement of a set of seed.

For example, consider to comprise for the profile contour of looking unit A and B of two connections and the concrete unified observability figure that blocks the border polyline of correspondence.How following table can generate or walk around increment G+ sub-grid or increment G-sub-grid with the mark contour of the selection of depending on the set/tabulation of seed triangle if showing.

Therefore, because increment G+(and increment G-) the sub-grid data can generate according to the mark profile contour of looking the unit transition for correspondence in working time, so do not need to store increment G+(and increment G-) the sub-grid data.Because compare increment G sub-grid data, mark profile contour information has less carrying cost usually, so the method can reduce the overall storage demand of outer deposit data fully.This is especially when setting up when selected profile contour is reused in the unit transition for some looking, and it high place is set up partly ought effectively dynamically blocking (EDO), corresponding to the high district of the visible continuity of the local dynamic station of model.

In order to generate for the increment G+ that looks the unit transition or increment G-sub-grid information in working time, importantly all polygonal mesh datas (looking as seen from the specific unit of looking) and mark profile contour information (corresponding to specifically looking the unit transition) have appeared at fast accessible (interior) storer.

In order to ensure before increment G sub-grid generate to be processed from this information of external memory storer prestrain, server process can directly be accessed from what comprise that will generate increment G sub-grid data specific look the unit and be looked more greatly the look set of visible polygon or polygon segmentation of unit.These data are equivalent to for the direct PVS data of looking the unit, but minimize storage demand by precomputation and storage for these direct PVS data of looking greatly the unit of relative few quantity, during looking ahead, uses the trellis traversal processing simplified and mark profile contour information to be paired in the more substantial involved PVS that looks the unit next life according to above-mentioned direct PVS data.Alternately, can before construct the father and looked the unit for the adjacent existing VM/PVS that constructs unit (self being that the father that other sons are looked the unit looks the unit) that looks greatly to father's the increment G sub-grid of looking the unit transition is data based for the father by access.

In addition, server process is also accessed the mark profile contour information of looking the unit for the specific son that will generate increment G sub-grid information.

When in internal storage, presenting structure increment G in the required polygonal mesh information of grid data and mark profile contour information, the generation of the increment G sub-grid data that the external memory memory access can not ISR.

It is to adopt the example that is present in the simple observability hierarchical structure between the hierarchy unit that the father is looked this purposes that unit and son look cell data: look unit all the polygon/segmentations of looking visible from son and also look the unit and look as seen from comprising the father.Look in this application of unit the father, importantly comprise in fact being positioned at all polygon/segmentations that the father looks the unit.Finally, usually the terminal leaf is looked unit structure and become to comprise seldom or do not comprise geometric figure, so that their VM/PVS " looks from looking the unit " exclusively and less " is comprised in and looks the unit ".

It is identical with the step 4620 shown in Figure 46 and higher step that step 5220 reaches higher step.

As previously mentioned, the increment PVS scheme of this method adopts and to look between the unit (for example, adjacent father looks the unit) and the father looks unit and the sub clear and definite increment visibility information of looking the unit transition of looking between the unit for adjacent.

In one embodiment, use clear and definite increment G+ and/or increment G-data acquisition for the father to father's the increment visibility information of looking the unit transition, and use middle mark profile contour information for the father to the unit transition of looking of son and/or son to son.This layout makes it possible to carry out balance in storage size and working time between the transmission/processing cost.The multi-level hierarchy of these layouts can further reduce carrying cost.

In addition, the client-server of Figure 52 is implemented to be obedient to following alternate embodiments: client process is presented on (similar and Figure 47) in the identical equipment with server process.Has effectiveness in the counting game of this embodiment in the external memory that whole database is stored on the computing equipment identical with the computing equipment of client and other application programs.

Figure 54 shows process flow diagram/block diagram that following client and server is processed: the server access pre-stored is looked the increment G sub-grid of unit and is configured to the observability figure that the father looks the unit/PVS with this information for the father, then the pre-stored of profile contour information use this observability figure/PVS to look the unit transition together with to(for) son is configured to the increment G sub-grid data that son is looked the unit transition, in this embodiment, the server increment G data that will comprise alternatively mark profile contour directly offer client.

Figure 54 shows and the similar a kind of embodiment of the embodiment of Figure 52.

Be labeled as that 5405 external memory district comprises increment G information and for the mark profile contour information of the increment G data of correspondence (the mark profile contour information that for example, is stored together with increment G+ data).

In the embodiment of Figure 54, alternatively, these data directly can be sent to Terminal Server Client and process.

In step 5440, client process is according to the increment G data configuration PVS data of correspondence.

In step 5450, client process uses the 3D trellis traversal processing (it has used included mark profile contour information) simplification, hint of Figure 36 and Figure 37 to come constructor PVS data according to father PVS data.

In addition, corresponding markers step and memory block and Figure 52's is identical.

The embodiment of Figure 54 can be less by transmitting with included mark profile contour information, relatively large increment G divides into groups to reduce transmission cost.In the method, client process structure less, son looks the PVS of unit, sends the correspondence of unit of the needs of increment G grouping look to(for) son thereby eliminated.In many cases, compare the correspondence of unit of increment G grouping look to(for) son, mark profile contour information has less storage and transmission cost.

In this embodiment, reduce transmission cost take the additional client runtime processing that generates sub-PVS data according to mark profile contour as cost.In the limited situation of available transmission bandwidth, can adopt the method to accelerate to transmit visibility events information to client.

Reuse the method for the mark profile contour of repeatedly looking the unit transition

In the district of the high observability continuity of having of model, can be identical or closely similar for some adjacent single order profile contours of looking the unit.If look the unit transition for several times, corresponding mark contour is identical, then can reuse simply the single marking contour.

When the mark contour is similar fully, storage and the most conservative mark contour of use, thus reduce storage and transmission cost.

Alternately, can encode to the difference between two profile contours (for example, by only the different ambiguity line segment of mark) and with this information be used for the working time/stage of looking ahead comes increment ground to construct a profile contour according to existing profile contour.

When having inadequate visibility events data, the prediction client uses the method for less increment observability grouping

Generally, large increment observability grouping is easy to reduce overall storage size, reduces repeating transmission and minimizing hunting time of element.This higher delay that is easy to transmit increment observability data is that cost improves handling capacity.On the contrary, the grouping of little increment observability generally have higher carrying cost, increase repeat transmit and higher hunting time, but can transmit them with lower delay.

In the client-server of this method is implemented, can require dynamically to select based on the transient delay of current visibility events stream the packet size of increment observability data.Distinguishingly, if prediction (by the speculative prefetching based on navigation of server) current view point can be transferred to client in the increment observability data of correspondence and penetrate in the past and specifically look the unit, then server can be with the increment visibility information corresponding with the transition of neglecting unit (for example, son is looked the unit).This grouping can be generated by server (for example, coming " walking around " specific polygon with the father to sub-mark profile contour) as required.Then the less increment observability grouping with correspondence sends to client.If transmission bandwidth is limited, then the method can reduce to transmit the time of grouping effectively.

The low visible continuous item that the method can be combined in roaming emphasizes that the method that can reduce level of detail during these time periods when connecting uses.

Improve the method for the robustness of implementing

The desired various polygon-polygons of this method determine that with other geometric figure cross-shaped portions the identical numerical accuracy that is subjected to run into limits in most of floating-point geometric figures calculate.One of these restrictions are relevant with the decidability of the cross-shaped portion of (for example, a polygonal limit is positioned at another polygonal plane and inside) under degenerate case.

Because this method is determined conservative result, so can increase a little certain little round values and effectively eliminate such degenerate case by looking the unit.Usually, this can remove and degenerate and the conservative solution of reservation.

By improve the method for the validity of enforcement with the level observability

Conservative linearization umbra uncontinuity grid, observability figure and the PAU that looks the unit from son looks the subset that the father of unit looks the linearization umbra uncontinuity grid of determining the unit according to comprising son.

Thereby, be used for determining that by any means of these methods looking the look visible collection of polygon/polygon segmentation of set of visible polygon/segmentation of unit from son can be limited to from comprising the father and look unit those polygons/polygon segmentation of looking visible.

By select to improve the method for observability precomputation efficient with shelter

After having constructed observability figure, use the method for the simplification of the previously described observability figure that measures to instruct the result with ESO.

Before the structure of observability figure, use the another kind of method of simplifying observability figure.In the method, use simple tolerance that the possibility of blocking of each mesh object (or the mesh object of closely separating grouping) is estimated.These tolerance can comprise the maximum subsystem of object or the bounding box of object.Alternately, tolerance can be to being formed on the estimation of the simple PAU between the bounding box of looking unit and mesh object.Use such estimation of blocking possibility of mesh object, can before structure observability figure, remove many invalid shelters.

Notice, select not explain that based on the shelter of the type of blocking possibility of each mesh object the larger polymerization that the larger PAU that produces causes blocks when being fused to other mesh object shelters by concrete mesh object shelter.Therefore, near checking other when the vicinity of shelter and position shelter select heuristics to become more effective.

Use is according to the method for the virtual shelter of PAU derivation

At structure shown in Figure 26 from the modification of the method for the PAU that looks the unit and look, with PAU and one or more Plane intersects.The cross-shaped portion on PAU and plane defines the virtual shelter of polygon.These virtual shelters can use to be similar to the mode of constructing the people such as Koltun (2000) of PVS in working time.

The advantage of this method is that it has realized that full 3D implements, and the people's such as Koltun (2000) method is 2.5D.

Using the method for the virtual shelter of polygon is the alternative method of following preceding method: mark profile contour (comprising the contour of conservatively simplifying) is used for delimiting virtual shelter effectively.

Obviously, in view of above teaching, can carry out various modifications and variations to this progress.Therefore, unless should be appreciated that in this article particularly description, otherwise can realize within the scope of the appended claims described progress.

Claims (155)

1. determine that described grid polygon forms polygonal mesh, said method comprising the steps of from the look method of set of visible grid polygon or the polygonal segmentation of described grid of the vision area with a plurality of vision areas summit for one kind:

A) determine at least one support polygon between described vision area and the described polygonal mesh;

B) construct at least one wedge according to described at least one support polygon, thereby described at least one wedge extends away from described vision area and exceeds described at least polygonal mesh;

C) determine the polygonal one or more cross-shaped portions of described wedge and described grid; And

D) determine from the look described set of visible described grid polygon or the polygonal segmentation of described grid of described vision area with described definite one or more cross-shaped portions of described at least one wedge and described polygonal mesh.

2. method according to claim 1 further comprises:

E) determine at least one single order silhouette edge of described polygonal mesh.

3. method according to claim 2, wherein,

Described at least one support polygon is to determine between at least one summit of described vision area and described at least one single order silhouette edge.

4. method according to claim 3, wherein, described at least one single order silhouette edge comprises:

Share the first polygon and second polygon of described at least one single order silhouette edge,

The described first polygon back side is to each the vision area summit from described a plurality of vision areas summit,

Described the second polygon is just being faced at least one the vision area summit from described a plurality of vision areas summit, and

Described the first polygon and described the second polygon relative to each other have the back side to orientation.

5. method according to claim 4, wherein, determine that the described step b) of at least one support polygon further comprises:

Determine to support the vision area summit for described at least one single order silhouette edge, described support vision area summit be restricted to from described a plurality of vision areas summit when the first vision area summit that runs into during towards described vision area Plane of rotation around described single order silhouette edge, the plane of described rotation is incident on the described single order silhouette edge and with respect to just not facing from each vision area summit on described a plurality of vision areas summit.

6. method according to claim 4, wherein, determine that the described step b) of at least one support polygon further comprises:

For the definite support vision area summit from described a plurality of vision areas summit of described at least one single order silhouette edge, described support vision area summit and described single order silhouette edge form the plane, described plane has identical side sexual orientation with described the first polygon of sharing described single order silhouette edge, and described plane is with respect to just not facing from each vision area summit on described a plurality of vision areas summit.

7. method according to claim 4, wherein,

When having determined that described at least one single order silhouette edge and another single order silhouette edge are shared the summit on the inside corner summit that forms described polygonal mesh, described at least one support polygon forms by described inside corner summit with from least two vision area summits on described a plurality of vision areas summit, described at least two vision area summits and described inside corner summit form the plane

Described plane has identical side sexual orientation with described the first polygon of sharing described at least one single order silhouette edge, and

Described plane is with respect to just not facing from each vision area summit on described a plurality of vision areas summit.

8. method according to claim 3, wherein,

When having determined that described at least one wedge intersects with described at least one the single order silhouette edge that forms the compound-contoured summit, described at least one support polygon forms by described compound-contoured summit with from least two vision area summits on described a plurality of vision areas summit, described at least two vision area summits and described inside corner summit form the plane, and described plane is with respect to just not facing from each vision area summit on described a plurality of vision areas summit.

9. method according to claim 5, wherein, the described step b) of described at least one wedge of structure is further comprising the steps:

B1) determine the plane of described at least one support polygon, and

B2) described at least one wedge of structure, described at least one wedge comprises the described plane of described at least one support polygon;

B3) by making at described constitution step b2) in described at least one wedge of forming intersect to limit described at least one wedge with at least the second wedge, described at least the second wedge and described at least one wedge are shared the polygonal summit of described grid.

10. method according to claim 6, wherein, the described step b) of described at least one wedge of structure is further comprising the steps:

B1) determine the plane of described at least one support polygon, and

B2) structure the first wedge, wherein, described the first wedge comprises the described plane of described at least one support polygon;

B3) by making at described constitution step b2) in described at least one wedge of forming intersect to limit described at least one wedge with at least the second wedge, wherein, described at least the second wedge and described at least one wedge are shared the polygonal described compound-contoured of described grid summit.

11. method according to claim 3, wherein, the described step c) of described at least one wedge of structure is further comprising the steps:

C1) construct the extension edge of described at least one support polygon by the limit of extending described at least one support polygon away from described vision area, intersect at least one summit of described extension edge and described at least one single order silhouette edge,

C2) construct described at least one wedge according to described at least one summit of described extension edge and described at least one single order silhouette edge.

12. method according to claim 1, wherein, described polygonal mesh comprises stream shape triangular mesh.

13. method according to claim 4, wherein, described at least one support polygon is conservative support polygon, and described conservative support polygon is to construct in the situation of not considering any polygonal mesh between two parties between described vision area and described at least one single order silhouette edge.

14. method according to claim 1, wherein, described polygonal mesh comprises non-convex polygon grid.

15. method according to claim 3 wherein, determines that the described step b) of at least one support polygon is further comprising the steps:

D) be configured to described polygonal mesh each tip, between described vision area and the described polygonal mesh at least one scan support polygon, described tip is restricted to the summit of single order silhouette edge, wherein, the described summit of single order profile is not the summit for any other single order silhouette edge.

16. method according to claim 9, wherein, described at least one single order silhouette edge is not the entrance limit.

17. method according to claim 10, wherein, described at least one single order silhouette edge is not the entrance limit.

18. method according to claim 5, wherein, comprise from the look described set of visible grid polygon or the polygonal segmentation of described grid of described vision area: from look visible at least one grid polygon and from described vision area at least one the grid polygon of looking sightless of described vision area.

19. method according to claim 4 wherein, determines that the described step a) of described at least one support polygon further comprises:

Determine high-order summit support polygon, wherein, described high-order summit support polygon is limited by described at least one single order silhouette edge and the visible vision area summit of supporting, described visible support vision area summit is restricted to the first visible vision area summit that runs into during towards described vision area Plane of rotation when around described single order silhouette edge, described Plane of rotation is incident on the described single order silhouette edge and with respect to each visible vision area summit and does not just face, and as seen look from described single order silhouette edge on described visible vision area summit.

20. method according to claim 19 wherein, determines that described at least one visible vision area summit further comprises:

Described at least one single order silhouette edge is used as the wire vision area;

At least one back projection single order silhouette edge of described polygonal mesh in the axle of determining between described wire vision area and described vision area, to form, described at least one back projection single order silhouette edge further comprises:

Share the first polygon and second polygon of described at least one back projection single order silhouette edge,

The described first polygon back side is to each wire vision area summit of described wire vision area,

Described the second polygon is just being faced at least one wire vision area summit of described wire vision area, and

Described the first polygon and described the second polygon relative to each other have the back side to orientation;

Determine at least one summit of described at least one back projection single order silhouette edge and at least one the back projection support polygon between the described wire vision area;

Construct at least one back projection wedge according to described at least one back projection support polygon;

Determine polygonal one or more cross-shaped portions of described at least one back projection wedge and described vision area; And

Determine described at least one visible vision area summit with described polygonal described one or more cross-shaped portions of described at least one back projection wedge and described vision area.

21. method according to claim 19 is further comprising the steps:

Determine the angle between the first plane and the second plane,

Described the first plane is the plane of supporting described at least one single order silhouette edge and described vision area,

Described the second plane is the plane that described at least one single order silhouette edge is separated with described vision area;

When having determined that described angle between described the first plane and described the second plane is greater than predetermined value, construct at least one wedge according to described high-order summit support polygon.

22. method according to claim 19, wherein, when having determined that at least one grid polygon between described at least one support polygon and described at least one single order silhouette edge and the described vision area intersects, carry out described definite high-order summit support polygon.

23. method according to claim 19, wherein, when having determined that the support polygon corresponding with at least one section of described single order silhouette edge and at least one grid polygon intersect, for the described at least one section described definite described high-order summit of the execution support polygon of described single order silhouette edge.

24. determine that described grid polygon forms polygonal mesh, said method comprising the steps of from the look method of set of visible grid polygon or described polygonal segmentation of the vision area with a plurality of summits for one kind:

A) determine a plurality of single order silhouette edges of described polygonal mesh, each single order silhouette edge is shared by the first polygon and second polygon of correspondence,

The described first polygon back side is to each vision area summit,

Described the second polygon is just being faced at least one vision area summit, and

Described the first polygon and described the second polygon relative to each other have the back side to orientation;

B) be identified for the summit support polygon of each determined single order silhouette edge, each determined summit support polygon limits by the single order silhouette edge of correspondence with from the support vision area summit of the correspondence on described a plurality of vision areas summit, described support vision area summit is restricted to the first vision area summit that runs into during towards described vision area rotary support plane when around the single order silhouette edge of described correspondence, and described supporting plane is incident on the described single order silhouette edge and with respect to just not facing from each vision area summit on described a plurality of vision areas summit;

C) be configured to the support summit wedge of each described summit support polygon by the limit of extending each corresponding summit support polygon away from described vision area, intersect on the first summit of the single order silhouette edge that described limit is corresponding with each and the second summit;

D) determine at least one limit support polygon, described at least one limit support polygon is limited by the inside corner summit of supporting vision area limit and one or more single order silhouette edges, described support vision area limit and described inside corner summit form the plane, and described plane is with respect to just not facing from each vision area summit on described a plurality of vision areas summit;

E) be configured to the support edge wedge of described at least one limit support polygon by the limit of extending described at least one the limit support polygon that intersects with described inside corner single order profile summit away from described vision area;

F) determine the support summit wedge of each described structure and one or more cross-shaped portions of described polygonal mesh, and determine the support edge wedge of described structure and one or more cross-shaped portions of described polygonal mesh; And

G) use described definite one or more cross-shaped portions of each described support summit wedge and described polygonal mesh and described definite one or more cross-shaped portions of described support edge wedge and described polygonal mesh, come to determine from the look set of described segmentation of visible described grid polygonal mesh or described polygonal mesh of described vision area.

25. determine that described grid polygon forms polygonal mesh, said method comprising the steps of from the look method of set of visible grid polygon or described polygonal segmentation of the vision area with a plurality of summits for one kind:

A) determine a plurality of single order silhouette edges of described polygonal mesh, each single order silhouette edge is shared by the first polygon and second polygon of correspondence,

The described first polygon back side is to each vision area summit,

Described the second polygon is just being faced at least one vision area summit, and

Described the first polygon and described the second polygon relative to each other have the back side to orientation;

B) be identified for the summit support polygon of each determined single order silhouette edge, each determined summit support polygon limits by the single order silhouette edge of correspondence with from the support vision area summit of the correspondence on described a plurality of vision areas summit, described support vision area summit is restricted to the first vision area summit that runs into during towards described vision area rotary support plane when around the single order silhouette edge of described correspondence, and described supporting plane is incident on the described single order silhouette edge and with respect to just not facing from each vision area summit on described a plurality of vision areas summit;

C) be configured to the support summit wedge of each described summit support polygon by the limit of extending each corresponding summit support polygon away from described vision area, intersect on the first summit of the single order silhouette edge that described limit is corresponding with each and the second summit;

D) determine at least one limit support polygon, described at least one limit support polygon is by the compound-contoured summit and support the vision area limit and limit, and described compound-contoured summit comprises at least one described support summit wedge and cross-shaped portion from least one described single order silhouette edge of described a plurality of single order silhouette edges;

E) be configured to the support edge wedge of described at least one limit support polygon by the limit of extending described at least one the limit support polygon that intersects with described compound-contoured summit away from described vision area;

F) determine the support summit wedge of each described structure and one or more cross-shaped portions of described polygonal mesh, and determine the support edge wedge of described structure and one or more cross-shaped portions of described polygonal mesh; And

G) determine from the look set of visible described grid polygonal mesh or the polygonal described segmentation of described grid of described vision area with described definite one or more cross-shaped portions of described definite one or more cross-shaped portions of each described support summit wedge and described polygonal mesh and described support edge wedge and described polygonal mesh.

26. method according to claim 25, wherein, described summit support polygon is the conservative support polygon of constructing in the polygonal situation of any grid between two parties of not considering between described vision area and described at least one single order silhouette edge, and described at least one limit support polygon is the conservative support polygon of constructing in the polygonal situation of any grid between two parties of not considering between described vision area and described at least one compound-contoured summit.

27. determine that described grid polygon forms polygonal mesh, said method comprising the steps of from the look method of set of visible polygon or described polygonal segmentation of the vision area with a plurality of summits for one kind:

A) determine a plurality of single order silhouette edges of described polygonal mesh, each single order silhouette edge is shared by the first polygon and second polygon of correspondence,

The described first polygon back side is to each vision area summit,

Described the second polygon is just being faced at least one vision area summit, and

Described the first polygon and described the second polygon relative to each other have the back side to orientation;

B) be identified for the summit support polygon of each determined single order silhouette edge, each determined summit support polygon limits by the single order silhouette edge of correspondence with from the support vision area summit of the correspondence on described a plurality of vision areas summit, described support vision area summit is restricted to the first vision area summit that runs into during towards described vision area rotary support plane when around the single order silhouette edge of described correspondence, and described supporting plane is incident on the described single order silhouette edge and with respect to just not facing from each vision area summit on described a plurality of vision areas summit;

C) be configured to the support summit wedge of each described summit support polygon, described structure further comprises:

C1) determine the plane of described summit support polygon,

C2) the support summit wedge of the described correspondence of structure, the support summit wedge of described correspondence comprises described plane,

C3) by making at described constitution step c2) in the support summit wedge of the described correspondence that forms with another supports summit wedge and intersects to limit described corresponding support summit wedge at least, it is described that another supports the shared polygonal summit of described grid of support summit wedge of summit wedge and described correspondence at least;

D) determine the support summit wedge of described structure and one or more cross-shaped portions of described polygonal mesh; And

E) determine from the look described set of described segmentation of visible described grid polygonal mesh or described polygonal mesh of described vision area with described definite one or more cross-shaped portions of described support summit wedge and described polygonal mesh.

28. method according to claim 27, wherein, described summit support polygon is the conservative support polygon of constructing in the polygonal situation of any grid between two parties of not considering between described vision area and described at least one single order silhouette edge.

29. method according to claim 24 wherein, is determined further to comprise from the look step g) of described set of visible described grid polygon or the polygonal segmentation of described grid of described vision area:

Determine to support summit wedge or support edge wedge and polygonal at least one cross-shaped portion of grid and whether block the border from what vision area was looked.

30. method according to claim 27, wherein, described determine described support summit wedge and described support edge wedge and polygonal at least one cross-shaped portion of grid whether from vision area look to block the border further comprising the steps:

Select the point on polygonal described at least one cross-shaped portion of described support summit wedge or described support edge wedge and grid,

Select the borderline point of described vision area, described point is positioned at not on the border on the limit of described vision area,

Determine the line segment between the described borderline described point of described point on polygonal described at least one cross-shaped portion of described support summit wedge or described support edge wedge and described grid and described vision area,

Determine the line segment intersection section of described line segment and described grid polygon and described support summit wedge or described support edge wedge; And

Whether determine to be positioned at described described borderline described point from described vision area on polygonal described at least one cross-shaped portion of described support summit wedge or described support edge wedge and described grid with the described line segment intersection section of described line segment looks and is blocked.

31. method according to claim 24, wherein,

At least one described support summit wedge and at least one described support edge wedge shape become to have the side of not being blocked and the side that is blocked block continuously boundary surface,

As seen the grid polygon that is positioned on the described described side that is not blocked of blocking continuously boundary surface looks from described vision area, and

The grid polygon that is positioned on the described described side that is blocked of blocking continuously boundary surface is looked invisible from described vision area.

32. method according to claim 25, wherein,

At least one described support summit wedge and at least one described support edge wedge shape become to have the side of not being blocked and the side that is blocked block continuously boundary surface,

As seen the grid polygon that is positioned on the described described side that is not blocked of blocking continuously boundary surface looks from described vision area, and

The grid polygon that is positioned on the described described side that is blocked of blocking continuously boundary surface is looked invisible from described vision area.

33. method according to claim 25, wherein, described definite at least one limit support polygon further comprises:

Determine to be in the first side support polygon at place, the first inside corner summit, described definite further comprising the steps:

Determine the angle between the first plane and the line,

Described the first plane is the plane of described first side support polygon;

Described line is the intersection on the second plane and the 3rd plane, described the second plane is the plane that is incident on the first summit support polygon on described the first inside corner summit, and described the 3rd plane is the plane that is incident on the second summit support polygon on described the first inside corner summit;

When having determined that described angle between described the first plane and the described line surpasses predetermined value, determine the described first side support polygon at place, described the first inside corner summit; And

When having determined that described angle between described the first plane and the described line surpasses described predetermined value, at structure support edge, described the first inside corner summit wedge, described support edge wedge comprise described the first summit support polygon and described the second summit support polygon on limited plane, described intersection place.

34. a reception is from the look method of set of visible grid polygon or the polygonal segmentation of described grid of the vision area with a plurality of vision areas summit, described grid polygon forms polygonal mesh, said method comprising the steps of:

A) receive from the look described set of visible grid polygon or the polygonal segmentation of described grid of described vision area, according to determine the polygonal described set of grid to get off:

The polygonal one or more cross-shaped portions of at least one wedge and described grid,

Described at least one wedge is incident on the single order silhouette edge; And

Described at least one wedge is to construct according at least one support polygon between described vision area and the described polygonal mesh; And

B) set of the described reception of look from described vision area visible grid polygon or the polygonal segmentation of described grid shown.

35. method according to claim 34, wherein, described single order silhouette edge comprises:

Share the first polygon and second polygon of described at least one single order silhouette edge,

The described first polygon back side is to each the vision area summit from described a plurality of vision areas summit,

Described the second polygon is just being faced at least one the vision area summit from described a plurality of vision areas summit, and

Described the first polygon and described the second polygon relative to each other have the back side to orientation.

36. method according to claim 35, wherein,

Described at least one support polygon is to determine between at least one summit of described vision area and described at least one single order silhouette edge.

37. method according to claim 36, wherein, determine to support the vision area summit for described single order silhouette edge, described support vision area summit be restricted to from described a plurality of vision areas summit when the first vision area summit that runs into during towards described vision area Plane of rotation around described single order silhouette edge, described Plane of rotation is incident on the described single order silhouette edge and with respect to just not facing from each vision area summit on described a plurality of vision areas summit.

38. method according to claim 36, wherein, determine to form with described single order silhouette edge the support vision area summit on plane, described plane has identical side sexual orientation with described the first polygon of sharing described single order silhouette edge, and described plane is with respect to just not facing from each vision area summit on described a plurality of vision areas summit.

39. method according to claim 35, wherein,

Described one or more cross-shaped portion comprises the cross-shaped portion of another wedge and described at least one wedge, and

Described another wedge is from being extended by another fixed support polygon of the margin of inside corner summit and described vision area, and described inside corner summit is formed by described single order silhouette edge and another single order silhouette edge.

40. method according to claim 35, wherein,

Described one or more cross-shaped portion comprises the cross-shaped portion of another wedge and described at least one wedge, and

Described another wedge is from being extended by another fixed support polygon of the margin of compound-contoured summit and described vision area, and described compound-contoured summit is formed by the cross-shaped portion between described at least one wedge and another single order silhouette edge.

41. determine that described grid polygon forms polygonal mesh from the look system of set of visible grid polygon or the polygonal segmentation of described grid of the vision area with a plurality of vision areas summit for one kind, described system comprises:

Scrambler with processor, described processor is configured to:

A) determine at least one support polygon between described vision area and the described polygonal mesh,

B) construct at least one wedge according to described at least one support polygon, thereby described at least one wedge extends away from described vision area and exceeds described at least polygonal mesh,

C) determine the polygonal one or more cross-shaped portions of described wedge and described grid, and

D) determine from the look described set of visible described grid polygon or the polygonal segmentation of described grid of described vision area with described definite one or more cross-shaped portions of described at least one wedge and described polygonal mesh; And

Client device, described client device shows the described set of look from described vision area visible grid polygon or the polygonal segmentation of described grid.

42. described system according to claim 41, wherein, the described processor of described scrambler further is configured to:

E) determine at least one single order silhouette edge of described polygonal mesh, wherein, described at least one support polygon is defined between at least one summit of described vision area and described at least one single order silhouette edge, and described at least one single order silhouette edge comprises:

Share the first polygon and second polygon of described at least one single order silhouette edge,

The described first polygon back side is to each the vision area summit from described a plurality of vision areas summit,

Described the second polygon is with respect to from just facing at least one vision area summit on described a plurality of vision areas summit, and

Described the first polygon and described the second polygon relative to each other have the back side to orientation.

43. described system according to claim 42, wherein, the described processor of described scrambler further is configured to:

Determine to support the vision area summit for described at least one single order silhouette edge, described support vision area summit be restricted to from described a plurality of vision areas summit when the first vision area summit that runs into during towards described vision area Plane of rotation around described single order silhouette edge, described Plane of rotation is incident on the described single order silhouette edge and with respect to just not facing from each vision area summit on described a plurality of vision areas summit, described at least one support polygon comprises described support vision area summit.

44. described system according to claim 43, wherein, the described processor of described scrambler further is configured to:

For the definite support vision area summit from described a plurality of vision areas summit of described at least one single order silhouette edge, described support vision area summit and described single order silhouette edge form the plane, and described plane has identical side sexual orientation with described the first polygon of sharing described single order silhouette edge, and described plane is not with respect to just facing from each vision area summit on described a plurality of vision areas summit, and described at least one support polygon comprises described support vision area summit.

45. described system according to claim 42, wherein, the described processor of described scrambler further is configured to:

Determine high-order summit support polygon, wherein, described high-order summit support polygon is limited by described at least one single order silhouette edge and the visible vision area summit of supporting, described visible support vision area summit is restricted to the first visible vision area summit that runs into during towards described vision area Plane of rotation when around described single order silhouette edge, described Plane of rotation is incident on the described single order silhouette edge and with respect to each visible vision area summit and does not just face, and as seen look from described single order silhouette edge on described visible vision area summit.

46. determine that described grid polygon forms polygonal mesh from the look system of set of visible grid polygon or described polygonal segmentation of the vision area with a plurality of summits for one kind, described system comprises:

Scrambler with processor, described processor is configured to:

A) determine a plurality of single order silhouette edges of described polygonal mesh, each single order silhouette edge is shared by the first polygon and second polygon of correspondence,

The described first polygon back side is to each vision area summit,

Described the second polygon is just being faced at least one vision area summit, and

Described the first polygon and described the second polygon relative to each other have the back side to orientation,

B) be identified for the summit support polygon of each determined single order silhouette edge, each determined summit support polygon limits by the single order silhouette edge of correspondence with from the support vision area summit of the correspondence on described a plurality of vision areas summit, described support vision area summit is restricted to the first vision area summit that runs into during towards described vision area rotary support plane when the single order silhouette edge around described correspondence, described supporting plane is incident on described single order silhouette edge and with respect to each the vision area summit from described a plurality of vision areas summit and does not just face

C) be configured to the support summit wedge of each described summit support polygon by the limit of extending each corresponding summit support polygon away from described vision area, intersect on the first summit of the single order silhouette edge that described limit is corresponding with each and the second summit,

D) determine at least one limit support polygon, described at least one limit support polygon is limited by the inside corner summit of supporting vision area limit and one or more single order silhouette edges, described support vision area limit and described inside corner summit form the plane, described plane is with respect to just not facing from each vision area summit on described a plurality of vision areas summit

E) be configured to the support edge wedge of described at least one limit support polygon by the limit of extending described at least one the limit support polygon that intersects with inside corner single order profile summit away from described vision area,

F) determine the support summit wedge of each described structure and one or more cross-shaped portions of described polygonal mesh, and determine the support edge wedge of described structure and one or more cross-shaped portions of described polygon wedge and described polygonal mesh, and

G) use described definite one or more cross-shaped portions of each described support summit wedge and described polygonal mesh and described definite one or more cross-shaped portions of described support edge wedge and described polygonal mesh, come to determine from the look described set of described segmentation of visible described grid polygonal mesh or described polygonal mesh of described vision area; And

Client device, described client device shows the described set of look from described vision area visible grid polygon or the polygonal segmentation of described grid.

47. determine that described grid polygon forms polygonal mesh from the look system of set of visible grid polygon or described polygonal segmentation of the vision area with a plurality of summits for one kind, described system comprises:

Scrambler with processor, described processor is configured to:

A) determine a plurality of single order silhouette edges of described polygonal mesh, each single order silhouette edge is shared by the first polygon and second polygon of correspondence,

The described first polygon back side is to each vision area summit,

Described the second polygon is just being faced at least one vision area summit, and

Described the first polygon and described the second polygon relative to each other have the back side to orientation,

B) be identified for the summit support polygon of each determined single order silhouette edge, each determined summit support polygon limits by the single order silhouette edge of correspondence with from the support vision area summit of the correspondence on described a plurality of vision areas summit, described support vision area summit is restricted to the first vision area summit that runs into during towards described vision area rotary support plane when the single order silhouette edge around described correspondence, described supporting plane is incident on described single order silhouette edge and with respect to each the vision area summit from described a plurality of vision areas summit and does not just face

C) be configured to the support summit wedge of each described summit support polygon by the limit of extending each corresponding summit support polygon away from described vision area, intersect on the first summit of the single order silhouette edge that described limit is corresponding with each and the second summit,

D) determine at least one limit support polygon, described at least one limit support polygon is limited by compound-contoured summit and support vision area limit, described compound-contoured summit comprises at least one described support summit wedge and cross-shaped portion from least one described single order silhouette edge of described a plurality of single order silhouette edges

E) be configured to the support edge wedge of described at least one limit support polygon by the limit of extending described at least one the limit support polygon that intersects with described compound-contoured summit away from described vision area,

F) determine the support summit wedge of each described structure and one or more cross-shaped portions of described polygonal mesh, and determine the support edge wedge of described structure and one or more cross-shaped portions of described polygonal mesh, and

G) use described definite one or more cross-shaped portions of each described support summit wedge and described polygonal mesh and described definite one or more cross-shaped portions of described support edge wedge and described polygonal mesh, come to determine from the look described set of visible described grid polygonal mesh or the polygonal described segmentation of described grid of described vision area;

Client device, described client device shows the described set of look from described vision area visible grid polygon or the polygonal segmentation of described grid.

48. determine that described grid polygon forms polygonal mesh from the look system of set of visible polygon or described polygonal segmentation of the vision area with a plurality of summits for one kind, described system comprises:

Scrambler with processor, described processor is configured to:

A) determine a plurality of single order silhouette edges of described polygonal mesh, each single order silhouette edge is shared by the first polygon and second polygon of correspondence,

The described first polygon back side is to each vision area summit,

Described the second polygon is just being faced at least one vision area summit, and

Described the first polygon and described the second polygon relative to each other have the back side to orientation,

B) be identified for the summit support polygon of each determined single order silhouette edge, each determined summit support polygon limits by the single order silhouette edge of correspondence with from the support vision area summit of the correspondence on described a plurality of vision areas summit, described support vision area summit is restricted to the first vision area summit that runs into during towards described vision area rotary support plane when the single order silhouette edge around described correspondence, described supporting plane is incident on described single order silhouette edge and with respect to each the vision area summit from described a plurality of vision areas summit and does not just face

C) be configured to the support summit wedge of each described summit support polygon, in order to construct described support summit wedge, described processing further is configured to:

C1) determine the plane of described summit support polygon,

C2) the support summit wedge of the described correspondence of structure, the support summit wedge of described correspondence comprises described plane,

C3) the support summit wedge by making described correspondence is with another supports summit wedge and intersects to limit described corresponding support summit wedge at least, and described at least another support summit wedge that supports summit wedge and described correspondence is shared the polygonal summit of described grid,

D) determine the support summit wedge of described structure and one or more cross-shaped portions of described polygonal mesh,

E) determine from the look described set of described segmentation of visible described grid polygonal mesh or described polygonal mesh of described vision area with described definite one or more cross-shaped portions of described support summit wedge and described polygonal mesh; And

Client device, described client device shows the described set of look from described vision area visible grid polygon or the polygonal segmentation of described grid.

49. non-transient state computer-readable recording medium with executable instruction stored thereon, when carrying out described executable instruction by processor, described executable instruction so that described processor determine from the look method of set of visible grid polygon or the polygonal segmentation of described grid of the vision area with a plurality of vision areas summit, described grid polygon forms polygonal mesh, said method comprising the steps of:

A) determine at least one support polygon between described vision area and the described polygonal mesh;

B) construct at least one wedge according to described at least one support polygon, thereby described at least one wedge extends away from described vision area and exceeds described at least polygonal mesh;

C) determine the polygonal one or more cross-shaped portions of described wedge and described grid; And

D) determine from the look described set of visible described grid polygon or the polygonal segmentation of described grid of described vision area with determined one or more cross-shaped portions of described at least one wedge and described polygonal mesh.

50. described non-transient state computer-readable medium according to claim 48, wherein, described executable instruction is further so that described processor execution in step:

E) determine at least one single order silhouette edge of described polygonal mesh, wherein,

Described at least one support polygon is to determine between at least one summit of described vision area and described at least one single order silhouette edge, and

Described at least one single order silhouette edge comprises:

Share the first polygon and second polygon of described at least one single order silhouette edge,

The described first polygon back side is to each the vision area summit from described a plurality of vision areas summit,

Described the second polygon is just being faced at least one the vision area summit from described a plurality of vision areas summit, and

Described the first polygon and described the second polygon relative to each other have the back side to orientation.

51. described non-transient state computer-readable medium according to claim 50, wherein, for the described step b) of determining at least one support polygon, described executable instruction is further so that described processor carries out step:

Determine to support the vision area summit for described at least one single order silhouette edge, described support vision area summit be restricted to from described a plurality of vision areas summit when the first vision area summit that runs into during towards described vision area Plane of rotation around described single order silhouette edge, described Plane of rotation is incident on the described single order silhouette edge and with respect to just not facing from each vision area summit on described a plurality of vision areas summit.

52. 1 described non-transient state computer-readable recording medium according to claim 5, wherein, for the described step b) of determining at least one support polygon, described executable instruction is further so that described processor carries out step:

For the definite support vision area summit from described a plurality of vision areas summit of described at least one single order silhouette edge, described support vision area summit and described single order silhouette edge form the plane, described plane has identical side sexual orientation with described the first polygon of sharing described single order silhouette edge, and described plane is with respect to just not facing from each vision area summit on described a plurality of vision areas summit.

53. described non-transient state computer-readable medium according to claim 50, wherein, for the described step a) of determining described at least one support polygon, described executable instruction is further so that described processor carries out step:

Determine high-order summit support polygon, wherein, described high-order summit support polygon is limited by described at least one single order silhouette edge and the visible vision area summit of supporting, described visible support vision area summit is restricted to the first visible vision area summit that runs into during towards described vision area Plane of rotation when around described single order silhouette edge, described Plane of rotation is incident on the described single order silhouette edge and with respect to each visible vision area summit and does not just face, and as seen look from described single order silhouette edge on described visible vision area summit.

54. non-transient state computer-readable recording medium with executable instruction stored thereon, when carrying out described executable instruction by processor, described executable instruction so that described processor determine from the look method of set of visible grid polygon or the polygonal segmentation of described grid of the vision area with a plurality of vision areas summit, described grid polygon forms polygonal mesh, said method comprising the steps of:

A) determine a plurality of single order silhouette edges of described polygonal mesh, each single order silhouette edge is shared by the first polygon and second polygon of correspondence,

The described first polygon back side is to each vision area summit,

Described the second polygon is just being faced at least one vision area summit, and

Described the first polygon and described the second polygon relative to each other have the back side to orientation;

B) be identified for the summit support polygon of each determined single order silhouette edge, each determined summit support polygon limits by the single order silhouette edge of correspondence with from the support vision area summit of the correspondence on described a plurality of vision areas summit, described support vision area summit is restricted to the first vision area summit that runs into when the single order silhouette edge around described correspondence is orientated described vision area rotary support plane, described supporting plane is incident on the described single order silhouette edge and with respect to just not facing from each vision area summit on described a plurality of vision areas summit;

C) be configured to the support summit wedge of each described summit support polygon by the limit of extending each corresponding summit support polygon away from described vision area, intersect on the first summit of the single order silhouette edge that described limit is corresponding with each and the second summit;

D) determine at least one limit support polygon, described at least one limit support polygon is limited by the inside corner summit of supporting vision area limit and one or more single order silhouette edges, described support vision area limit and described inside corner summit form the plane, and described plane is with respect to just not facing from each vision area summit on described a plurality of vision areas summit;

E) be configured to the support edge wedge of described at least one limit support polygon by the limit of extending described at least one the limit support polygon that intersects with described inside corner single order profile summit away from described vision area;

F) determine the support summit wedge of each described structure and one or more cross-shaped portions of described polygonal mesh, and determine the support edge wedge of described structure and one or more cross-shaped portions of described polygonal mesh; And

G) use described definite one or more cross-shaped portions of each described support summit wedge and described polygonal mesh and described definite one or more cross-shaped portions of described support edge wedge and described polygonal mesh, come to determine from the look described set of described segmentation of visible described grid polygonal mesh or described polygonal mesh of described vision area.

55. non-transient state computer-readable recording medium with executable instruction stored thereon, when carrying out described executable instruction by processor, described executable instruction so that described processor determine from the look method of set of visible grid polygon or the polygonal segmentation of described grid of the vision area with a plurality of vision areas summit, described grid polygon forms polygonal mesh, said method comprising the steps of:

A) determine a plurality of single order silhouette edges of described polygonal mesh, each single order silhouette edge is shared by the first polygon and second polygon of correspondence,

The described first polygon back side is to each vision area summit,

Described the second polygon is just being faced at least one vision area summit, and

Described the first polygon and described the second polygon relative to each other have the back side to orientation;

B) be identified for the summit support polygon of each determined single order silhouette edge, each determined summit support polygon limits by the single order silhouette edge of correspondence with from the support vision area summit of the correspondence on described a plurality of vision areas summit, described support vision area summit is restricted to the first vision area summit that runs into during towards described vision area rotary support plane when around the single order silhouette edge of described correspondence, and described supporting plane is incident on the described single order silhouette edge and with respect to just not facing from each vision area summit on described a plurality of vision areas summit;

C) be configured to the support summit wedge of each described summit support polygon by the limit of extending each corresponding summit support polygon away from described vision area, intersect on the first summit of the single order silhouette edge that described limit is corresponding with each and the second summit;

D) determine at least one limit support polygon, described at least one limit support polygon is by the compound-contoured summit and support the vision area limit and limit, and described compound-contoured summit comprises at least one described support summit wedge and cross-shaped portion from least one described single order silhouette edge of described a plurality of single order silhouette edges;

E) be configured to the support edge wedge of described at least one limit support polygon by the limit of extending described at least one the limit support polygon that intersects with described compound-contoured summit away from described vision area;

F) determine the support summit wedge of each described structure and one or more cross-shaped portions of described polygonal mesh, and determine the support edge wedge of described structure and one or more cross-shaped portions of described polygonal mesh; And

G) determine from the look described set of visible described grid polygonal mesh or the polygonal described segmentation of described grid of described vision area with described definite one or more cross-shaped portions of described definite one or more cross-shaped portions of each described support summit wedge and described polygonal mesh and described support edge wedge and described polygonal mesh.

56. non-transient state computer-readable recording medium with executable instruction stored thereon, when carrying out described executable instruction by processor, described executable instruction so that described processor determine from the look method of set of visible grid polygon or the polygonal segmentation of described grid of the vision area with a plurality of vision areas summit, described grid polygon forms polygonal mesh, said method comprising the steps of:

A) determine a plurality of single order silhouette edges of described polygonal mesh, each single order silhouette edge is shared by the first polygon and second polygon of correspondence,

The described first polygon back side is to each vision area summit,

Described the second polygon is just being faced at least one vision area summit, and

Described the first polygon and described the second polygon relative to each other have the back side to orientation;

B) be identified for the summit support polygon of each determined single order silhouette edge, each determined summit support polygon limits by the single order silhouette edge of correspondence with from the support vision area summit of the correspondence on described a plurality of vision areas summit, described support vision area summit is restricted to the first vision area summit that runs into during towards described vision area rotary support plane when around the single order silhouette edge of described correspondence, and described supporting plane is incident on the described single order silhouette edge and with respect to just not facing from each vision area summit on described a plurality of vision areas summit;

C) be configured to the support summit wedge of each described summit support polygon, described structure further comprises:

C1) determine the plane of described summit support polygon,

C2) the support summit wedge of the described correspondence of structure, the support summit wedge of described correspondence comprises described plane,

C3) the support summit wedge by making the described correspondence that forms in described constitution step c2 is with another supports summit wedge and intersects to limit described corresponding support summit wedge at least, and described at least another support summit wedge that supports summit wedge and described correspondence is shared the polygonal summit of described grid;

D) determine the support summit wedge of described structure and one or more cross-shaped portions of described polygonal mesh; And

E) determine from the look described set of described segmentation of visible described grid polygonal mesh or described polygonal mesh of described vision area with described definite one or more cross-shaped portions of described support summit wedge and described polygonal mesh.

57. determine that described grid polygon forms polygonal mesh, said method comprising the steps of from the look method of set of visible grid polygon or the polygonal segmentation of described grid of vision area for one kind:

A) at least one that determine to be incident on the described grid polygon blocked the border from what vision area was looked, described at least one the border has the side of not being blocked and the side that is blocked from blocking of looking of vision area;

B) determine to add to from the look polygonal quantity of grid of described set of visible grid polygon or the polygonal segmentation of described grid of described vision area, described grid polygon is by at polygonal mesh and described at least one cross-shaped portion on the described side that is not blocked between the border of blocking of looking from vision area the face of described crossing polygonal mesh being repartitioned to add;

C) determine by from described at least one block the polygonal quantity of the grid that blocks on the border from what vision area was looked;

D) when having determined that the polygonal described quantity of the grid that adds surpasses the polygonal described quantity predetermined threshold of grid be blocked, remove described at least one block the border from what vision area was looked.

58. determine that described grid polygon forms polygonal mesh, said method comprising the steps of from the look method of set of visible grid polygon or the polygonal segmentation of described grid of vision area for one kind:

A) at least one that determine to be incident on the described grid polygon blocked the border from what vision area was looked;

B) determine by described at least one block the polygonal surface area of the grid that blocks on the border from what vision area was looked;

C) if the polygonal described surface area of the grid that is blocked less than predetermined value, then remove described at least one block the border from what vision area was looked.

59. determine that described grid polygon forms polygonal mesh, said method comprising the steps of from the look method of set of visible grid polygon or the polygonal segmentation of described grid of vision area for one kind:

A) at least one that determine to be incident on the described grid polygon blocked the border from what vision area was looked, described at least one have at least one limit from the border of blocking that vision area is looked;

B) determine to add to from the look polygonal quantity of grid of described set of visible grid polygon or the polygonal segmentation of described grid of vision area, described grid polygon is by at polygonal mesh and described at least one cross-shaped portion that blocks between the border of looking from vision area the face of described crossing polygonal mesh being repartitioned to add;

C) determine by described at least one block the polygonal quantity of the grid that blocks on the border from what vision area was looked;

D) when having determined that the polygonal described quantity of the grid that adds surpasses the polygonal described quantity predetermined threshold of grid that is blocked, construct described at least one simple version that blocks the border of looking from vision area, what described structure was included in that described at least one side structure of being blocked that blocks the border of looking from vision area revises blocks the border from what vision area was looked, described modification from the quantity on the limit of blocking the border that vision area the is looked quantity less than described at least one limit of blocking the border of looking from the district;

E) repartition described grid polygon with described at least one described simple version that blocks the border of looking from vision area.

60. determine that described grid polygon forms polygonal mesh, said method comprising the steps of from the look method of set of visible grid polygon or the polygonal segmentation of described grid of vision area for one kind:

A) at least one that determine to be incident on the described grid polygon blocked the border from what vision area was looked, and described at least one border of blocking of looking from the district has at least one limit;

B) to by the face of described crossing polygonal mesh repartitioned to add to from the look polygonal quantity of grid of described set of visible grid polygon or the polygonal segmentation of described grid of vision area determining at polygonal mesh and described at least one cross-shaped portion between the border of blocking of looking from the district;

C) determine by described at least one block the polygonal surface area of the grid that blocks on the border from what vision area was looked;

D) when having determined that the polygonal described quantity of the grid that adds surpasses the first predetermined value or the polygonal described surface area of grid that is blocked during less than second value of being scheduled to, construct described at least one simple version that blocks the border of looking from vision area, what described structure was included in that described at least one side structure of being blocked that blocks the border of looking from vision area revises blocks the border from what vision area was looked, described modification from the quantity on the limit of blocking the border that vision area is looked less than described at least one quantity from the limit of blocking the border that vision area is looked;

E) repartition described grid polygon with described at least one described simple version that blocks the border of looking from vision area.

61. a reception is from the look method of set of visible grid polygon or the polygonal segmentation of described grid of the vision area with a plurality of vision areas summit, described grid polygon forms polygonal mesh, said method comprising the steps of:

A) receive from the look set of visible grid polygon or the polygonal segmentation of described grid of described vision area,

The described set of polygonal mesh does not comprise that at least one that remove block the border from what vision area was looked from the described set of look from described vision area visible grid polygon or the polygonal segmentation of described grid, described at least one have the side of not being blocked and the side that is blocked from the border of blocking that vision area is looked

When the polygonal quantity of grid of the described set of having determined to add to grid polygon or the polygonal segmentation of described grid surpass by described at least one look from the district block the polygonal quantity predetermined threshold of the grid that blocks on the border time, from the described set of grid polygon or the polygonal segmentation of described grid, remove described at least one block the border, and

The polygonal described quantity of the grid that adds is by coming the face of described crossing polygonal mesh is repartitioned to add at polygonal mesh and described at least one cross-shaped portion on the described side that is not blocked between the border of blocking of looking from vision area; And

B) the described set of look from described vision area visible grid polygon or the polygonal segmentation of described grid shown.

62. a reception is from the look method of set of visible grid polygon or the polygonal segmentation of described grid of the vision area with a plurality of vision areas summit, described grid polygon forms polygonal mesh, said method comprising the steps of:

A) receive from the look described set of visible grid polygon or the polygonal segmentation of described grid of described vision area,

The polygonal described set of grid does not comprise that at least one that remove from the described set of look from described vision area visible grid polygon or the polygonal segmentation of described grid block the border from what vision area was looked, described at least one have the side of not being blocked and the side that is blocked from the border of blocking that vision area is looked

When determined by described at least one from vision area look block the polygonal surface area of the grid that blocks on the border less than predetermined value the time, from the described set of look from described vision area visible grid polygon or the polygonal segmentation of described grid, remove described at least one block the border; And

B) the described set of look from described vision area visible grid polygon or the polygonal segmentation of described grid shown.

63. determine that described grid polygon forms polygonal mesh from the look system of set of visible grid polygon or the polygonal segmentation of described grid of vision area for one kind, described system comprises:

Scrambler with processor, described processor is configured to:

A) at least one that determine to be incident on the described grid polygon blocked the border from what vision area was looked, described at least one the border has the side of not being blocked and the side that is blocked from blocking of looking of vision area;

B) determine to add to from the look polygonal quantity of grid of described set of visible grid polygon or the polygonal segmentation of described grid of described vision area, described grid polygon is by at polygonal mesh and described at least one cross-shaped portion on the described side that is not blocked between the border of blocking of looking from vision area the face of described crossing polygonal mesh being repartitioned to add;

C) determine by described at least one block the polygonal quantity of the grid that blocks on the border from what vision area was looked, and

D) when having determined that the polygonal described quantity of the grid that adds surpasses the polygonal described quantity predetermined threshold of grid be blocked, remove described at least one block the border from what vision area was looked; And

Client device, described client device shows the described set of look from described vision area visible grid polygon or the polygonal segmentation of described grid.

64. determine that described grid polygon forms polygonal mesh from the look system of set of visible grid polygon or the polygonal segmentation of described grid of vision area for one kind, described system comprises:

Scrambler with processor, described processor is configured to:

A) at least one that determine to be incident on the described grid polygon blocked the border from what vision area was looked;

B) determine by described at least one block the polygonal surface area of the grid that blocks on the border from what vision area was looked;

C) if the polygonal described surface area of the grid that is blocked less than predetermined value, then remove described at least one block the border from what vision area was looked; And

Client device, described client device shows the described set of look from described vision area visible described grid polygon or the polygonal segmentation of described grid.

65. determine that described grid polygon forms polygonal mesh from the look system of set of visible grid polygon or the polygonal segmentation of described grid of vision area for one kind, described system comprises:

Scrambler with processor, described processor is configured to:

A) determine to be incident on described grid polygonal at least one block the border from what vision area was looked, described at least one border of blocking of looking from the district has at least one limit;

B) determine to add to the polygonal quantity of grid of the described set of grid polygon or the polygonal segmentation of described grid, described grid polygon is by at polygonal mesh and described at least one cross-shaped portion that blocks between the border of looking from vision area the face of described crossing polygonal mesh being repartitioned to add;

C) determine by described at least one block the polygonal quantity of the grid that blocks on the border from what vision area was looked;

D) when having determined that the polygonal described quantity of the grid that adds surpasses the polygonal described quantity predetermined threshold of grid that is blocked, construct described at least one simple version that blocks the border of looking from vision area, what described structure was included in that described at least one side structure of being blocked that blocks the border of looking from vision area revises blocks the border from what vision area was looked, described modification from the quantity on the limit of blocking the border that vision area the is looked quantity less than described at least one limit of blocking the border of looking from the district;

E) repartition described grid polygon with described at least one described simple version that blocks the border of looking from vision area; And

Client device, described client device shows the described set of look from described vision area visible grid polygon or the polygonal segmentation of described grid.

66. determine that described grid polygon forms polygonal mesh from the look system of set of visible grid polygon or the polygonal segmentation of described grid of vision area for one kind, described system comprises:

Scrambler with processor, described processor is configured to:

A) at least one that determine to be incident on the described grid polygon blocked the border from what vision area was looked, and described at least one border of blocking of looking from the district has at least one limit;

B) determine by at polygonal mesh and described at least one cross-shaped portion that blocks between the border of looking from the district face of described crossing polygonal mesh being repartitioned to add to from the look polygonal quantity of grid of described set of visible grid polygon or the polygonal segmentation of described grid of vision area;

C) determine by described at least one block the polygonal surface area of the grid that blocks on the border from what vision area was looked;

D) when determined that the polygonal described quantity of the grid that adds surpasses the first predetermined value or the polygonal described surface area of grid that is blocked less than second value of being scheduled to, then construct described at least one simple version that blocks the border of looking from vision area, what described structure was included in that described at least one side structure of being blocked that blocks the border of looking from vision area revises blocks the border from what vision area was looked, described modification from the quantity on the limit of blocking the border that vision area is looked less than described at least one quantity from the limit of blocking the border that vision area is looked;

E) repartition described grid polygon with described at least one described simple version that blocks the border of looking from vision area; And

Client device, described client device shows the described set of the polygonal segmentation of the polygonal or described grid of visible grid of looking from described vision area.

67. non-transient state computer-readable recording medium with executable instruction stored thereon, when carrying out described executable instruction by processor, described executable instruction so that described processor determine from the look method of set of visible grid polygon or the polygonal segmentation of described grid of vision area, described grid polygon forms polygonal mesh, said method comprising the steps of:

A) at least one that determine to be incident on the described grid polygon blocked the border from what vision area was looked, described at least one the border has the side of not being blocked and the side that is blocked from blocking of looking of vision area;

B) determine to add to from the look polygonal quantity of grid of described set of visible grid polygon or the polygonal segmentation of described grid of described vision area, described grid polygon is by at polygonal mesh and described at least one phase cross-shaped portion on the described side that is not blocked between the border of blocking of looking from vision area the face of described crossing polygonal mesh being repartitioned to add;

C) determine by described at least one block the polygonal quantity of the grid that blocks on the border from what vision area was looked;

D) when having determined that the polygonal described quantity of the grid that adds surpasses the polygonal described quantity predetermined threshold of grid be blocked, remove described at least one block the border from what vision area was looked.

68. non-transient state computer-readable recording medium with executable instruction stored thereon, when carrying out described executable instruction by processor, described executable instruction so that described processor determine from the look method of set of visible grid polygon or the polygonal segmentation of described grid of vision area, described grid polygon forms polygonal mesh, said method comprising the steps of:

A) at least one that determine to be incident on the described grid polygon blocked the border from what vision area was looked;

B) determine by described at least one block the polygonal surface area of the grid that blocks on the border from what vision area was looked;

C) if the polygonal described surface area of the grid that is blocked less than predetermined value, then remove described at least one block the border from what vision area was looked.

69. non-transient state computer-readable recording medium with executable instruction stored thereon, when carrying out described executable instruction by processor, described executable instruction so that described processor determine from the look method of set of visible grid polygon or the polygonal segmentation of described grid of vision area, described grid polygon forms polygonal mesh, said method comprising the steps of:

A) at least one that determine to be incident on the described grid polygon blocked the border from what vision area was looked, described at least one have at least one limit from the border of blocking that vision area is looked;

B) determine to add to the polygonal quantity of grid of the described set of grid polygon or the polygonal segmentation of described grid, described grid polygon is by at polygonal mesh and described at least one cross-shaped portion that blocks between the border of looking from vision area the face of described crossing polygonal mesh being repartitioned to add;

C) determine by described at least one block the polygonal quantity of the grid that blocks on the border from what vision area was looked;

D) when having determined that the polygonal described quantity of the grid that adds surpasses the polygonal described quantity predetermined threshold of grid that is blocked, construct described at least one simple version that blocks the border of looking from vision area, what described structure was included in that described at least one side structure of being blocked that blocks the border of looking from vision area revises blocks the border from what vision area was looked, described modification from the quantity on the limit of blocking the border that vision area the is looked quantity less than described at least one limit of blocking the border of looking from the district;

E) repartition described grid polygon with described at least one described simple version that blocks the border of looking from vision area.

70. non-transient state computer-readable recording medium with executable instruction stored thereon, when carrying out described executable instruction by processor, described executable instruction so that described processor determine from the look method of set of visible grid polygon or the polygonal segmentation of described grid of vision area, described grid polygon forms polygonal mesh, said method comprising the steps of:

A) at least one that determine to be incident on the described grid polygon blocked the border from what vision area was looked, and described at least one border of blocking of looking from the district has at least one limit;

B) determine by at polygonal mesh and described at least one cross-shaped portion that blocks between the border of looking from vision area the face of described crossing polygonal mesh being repartitioned to add to from the look polygonal quantity of grid of described set of visible grid polygon or the polygonal segmentation of described grid of vision area;

C) determine by described at least one block the polygonal surface area of the grid that blocks on the border from what vision area was looked;

D) when determined that the polygonal described quantity of the grid that adds surpasses the first predetermined value or the polygonal described surface area of grid that is blocked less than second value of being scheduled to, then construct described at least one simple version that blocks the border of looking from vision area, what described structure was included in that described at least one side structure of being blocked that blocks the border of looking from vision area revises blocks the border from what vision area was looked, described modification from the quantity on the limit of blocking the border that vision area is looked less than described at least one quantity from the limit of blocking the border that vision area is looked;

E) repartition described grid polygon with described at least one described simple version that blocks the border of looking from vision area.

71. determine that described grid polygon forms polygonal mesh, said method comprising the steps of from the look method of set of visible grid polygon or the polygonal segmentation of described grid of the vision area with a plurality of summits for one kind:

At least one grid polygon from the polygonal described set of grid is traveled through;

When running at least one single order silhouette edge, end described traversal;

At described at least one wedge of single order silhouette edge structure;

Determine described at least one wedge with from polygonal at least one cross-shaped portion of another grid in the polygonal described set of grid;

Determine that whether described at least one cross-shaped portion blocks the border from what vision area was looked;

When determined described at least one cross-shaped portion be from vision area look block the border time continue described traversal in the described side that is not blocked of blocking the border of looking from vision area.

72. method according to claim 32 wherein, is carried out described step a), b), c), d) and order e) by following definite:

Travel through the described grid polygon of described polygonal mesh;

For each grid polygon that travels through, determine the grid polygon that in the grid polygon that is formed on the first traversal and the axle between the described vision area, whether exists at least one not travel through;

When having determined in described axle, to have the grid polygon of any not traversal, end described traversal at the grid polygon place of described the first traversal, and described traversal is restarted at described at least one the grid polygon place that does not travel through in described axle;

For each grid polygon that travels through, determine whether the grid polygon of described traversal intersects with the border of blocking from vision area is looked of storage;

When the grid polygon of having determined described traversal and describedly block the border when intersecting from what vision area was looked, end described traversal at the described boundary that blocks of looking from vision area;

When the grid polygon of having determined described traversal has limit as the single order silhouette edge, determine described vision area and as at least one support polygon between the described limit of single order silhouette edge;

Construct at least one wedge according to described at least one support polygon;

Determine described at least one wedge and the described not polygonal one or more cross-shaped portions of grid of traversal;

For each cross-shaped portion in described at least one wedge and the polygonal described one or more cross-shaped portions of described grid, determine that whether described cross-shaped portion blocks the border from what vision area was looked; And

Storage and described grid polygon intersect describedly blocks the border from what vision area was looked, wherein,

Comprise the grid polygon of described traversal from the look described set of visible grid polygon or the polygonal segmentation of described grid of described vision area.

73. 2 described methods according to claim 7, wherein, described polygonal mesh is by surrounding around polygonal mesh, and wherein, and described traversal around polygonal mesh comprises and surrounds the described around the traversal on the polygonal mesh side of described polygonal mesh.

74. 2 described methods according to claim 7, wherein, described traversal comprises breadth first traversal.

75. determine that described grid polygon forms polygonal mesh, said method comprising the steps of from the look method of set of visible grid polygon or the polygonal segmentation of described grid of the vision area with a plurality of summits for one kind:

Traversal is from least one grid polygon of the polygonal described set of grid;

When running at least one single order silhouette edge, end described traversal;

At described at least one wedge of single order silhouette edge structure;

Determine described at least one wedge with from polygonal at least one cross-shaped portion of another grid of the polygonal described set of grid;

Determine that whether described at least one cross-shaped portion is to block the border from what vision area was looked;

When determined described at least one cross-shaped portion be from vision area look block the border time, continue described traversal in the described side that is not blocked of blocking the border of looking from vision area.

76. a reception is from the look method of set of visible grid polygon or the polygonal segmentation of described grid of the vision area with a plurality of summits, described grid polygon forms polygonal mesh, said method comprising the steps of:

A) receive at the client device place from the look described set of visible grid polygon or the polygonal segmentation of described grid of described vision area, described set by scrambler according to comprising that following traversal method determines:

Traversal is from least one grid polygon of the polygonal described set of grid,

When running at least one single order silhouette edge, end described traversal,

At described at least one wedge of single order silhouette edge structure;

Determine described at least one wedge with from polygonal at least one cross-shaped portion of another grid in the polygonal described set of grid;

Determine that whether described at least one cross-shaped portion blocks the border from what vision area was looked;

When determined described at least one cross-shaped portion be from vision area look block the border time, continue described traversal in the described side that is not blocked of blocking the border of looking from vision area; And

B) show by the described set of described client device to look from described vision area visible grid polygon or the polygonal segmentation of described grid.

77. determine that described grid polygon forms polygonal mesh from the look system of set of visible grid polygon or the polygonal segmentation of described grid of the vision area with a plurality of summits for one kind, described system comprises:

Scrambler with processor, described processor is configured to:

At least one grid polygon from the polygonal described set of grid is traveled through;

When running at least one single order silhouette edge, end described traversal;

At described at least one wedge of single order silhouette edge structure;

Determine described at least one wedge with from polygonal at least one cross-shaped portion of another grid in the polygonal described set of grid;

Determine that whether described at least one cross-shaped portion blocks the border from what vision area was looked; And

When determined described at least one cross-shaped portion be from vision area look block the border time, continue described traversal in the described side that is not blocked of blocking the border of looking from vision area; And

Client device, described client device shows the set of look from vision area visible grid polygon or the polygonal segmentation of described grid.

78. non-transient state computer-readable recording medium with executable instruction stored thereon, when carrying out described executable instruction by processor, described executable instruction so that described processor determine from the look method of set of visible grid polygon or the polygonal segmentation of described grid of the vision area with a plurality of summits, described grid polygon forms polygonal mesh, said method comprising the steps of:

At least one grid polygon from the polygonal described set of grid is traveled through;

When running at least one single order silhouette edge, end described traversal;

At described at least one wedge of single order silhouette edge structure;

Determine described at least one wedge with from polygonal at least one cross-shaped portion of another grid in the polygonal described set of grid;

Determine that whether described at least one cross-shaped portion blocks the border from what vision area was looked;

When determined described at least one cross-shaped portion be from vision area look block the border time, continue described traversal in the described side that is not blocked of blocking the border of looking from vision area.

79. the method determining and store the information that first group of grid polygon is described, described grid polygon forms polygonal mesh, and as seen described first group of grid polygon look from the first vision area, said method comprising the steps of:

Determine and storage from the second vision area second group of grid polygon of looking visible, described the second vision area comprises described the first vision area;

Determine and the polygonal information of storage at least one grid of expression, described at least one grid polygon belongs to the polygonal set of grid of connection, and as seen the polygonal described set of the grid of connection looks from described the first vision area; And

Determine and store the information of at least one boundary edge of polygonal described at least one set of grid that expression connects.

80. 9 described methods according to claim 7, wherein, the described information that represents at least one boundary edge comprises the information that represents at least one single order silhouette edge, and described at least one single order silhouette edge is the single order silhouette edge with respect to described the first vision area, and described at least one single order silhouette edge comprises:

Share the first polygon and second polygon of described at least one single order silhouette edge,

Described the first polygon is with respect to each back side, vision area summit of described the first vision area pair,

Described the second polygon is just faced with respect at least one vision area summit of described the first vision area, and

Described the first polygon and described the second polygon relative to each other have the back side to orientation.

81. 9 described methods according to claim 7, wherein, the described information that represents at least one boundary edge comprises the information that represents at least one compound-contoured summit, described at least one compound-contoured summit is the compound-contoured summit with respect to described the first vision area, described compound-contoured summit comprises that described at least one single order silhouette edge comprises by the wedge of described the first vision area support and the cross-shaped portion between at least one single order silhouette edge:

Share the first polygon and second polygon of described at least one single order silhouette edge,

The described first polygon back side is to each vision area summit of described the first vision area,

Described the second polygon is just being faced at least one vision area summit of described the first vision area, and

Described the first polygon and described the second polygon relative to each other have the back side to orientation.

82. 9 described methods according to claim 7, wherein, the described information that represents at least one boundary edge comprises that at least one blocks boundary edge from what vision area was looked, described at least one boundary edge of blocking of looking from vision area is the limit that is formed by at least one wedge that supported by described the first vision area and the polygonal cross-shaped portion of described grid, described at least one wedge is the wedge that is incident on the single order silhouette edge, described single order silhouette edge is the single order silhouette edge with respect to described the first vision area, and described single order silhouette edge comprises:

Share the first polygon and second polygon of described single order silhouette edge,

The described first polygon back side is to each vision area summit of described the first vision area,

Described the second polygon is just being faced at least one vision area summit of described the first vision area, and

Described the first polygon and described the second polygon relative to each other have the back side to orientation.

83. 0 described method according to claim 8, wherein, describe the information that first group of polygonal described information of grid further comprises the length of the sequence that represents a plurality of single order silhouette edges, the described sequence of described a plurality of single order silhouette edges comprises the single order silhouette edge of at least summit-connection at least one summit that is connected to described at least one single order silhouette edge.

84. 9 described methods according to claim 7, wherein, the described information that represents at least one boundary edge comprises the polygonal information of at least one grid of expression, stops the breadth first traversal to the polygonal described set of grid that connects at described at least one grid polygon place.

85. 4 described methods according to claim 8, wherein, being illustrated in its place stops described information at least one grid polygon limit of the breadth first traversal of the polygonal described set of grid that connects and comprises and be stored to a few limit-run into numeral, described limit-run into numeral is the iteration number that runs into the required described breadth first traversal in described at least one grid polygon limit, and the described iteration of described breadth first traversal begins from described at least one grid polygon of the polygonal described set of grid that connects.

86. one kind is used institute's canned data to determine first group of polygonal method of grid, described grid polygon forms polygonal mesh, looks as seen from the first vision area from the polygonal first group of grid polygon of described grid, and described method comprises:

The visible second group of grid polygon of looking from the second vision area retrieved, and described the second vision area comprises described the first vision area;

To representing that the polygonal information of at least one the first grid retrieves, described at least one first grid polygon belongs to the polygonal set of grid of connection, and as seen the polygonal described set of the grid of connection looks from described the first vision area;

The information of at least one boundary edge of polygonal described at least one set of grid that expression is connected is retrieved;

Polygonal described at least one set travels through to the grid that belongs to described second group of polygonal connection of grid, and described traversal comprises:

Begin described traversal at described at least one first grid polygon place, and

When during described traversal, running into described at least one boundary edge, interrupt described traversal.

87. 6 described methods according to claim 8, wherein, the described information of described at least one boundary edge of expression comprises the information that represents at least one single order silhouette edge, described at least one single order silhouette edge is the single order silhouette edge with respect to described the first vision area, and described at least one single order silhouette edge comprises:

Share the first polygon and second polygon of described at least one single order silhouette edge,

The described first polygon back side is to each the vision area summit from described a plurality of vision areas summit,

Described the second polygon is just being faced at least one the vision area summit from described a plurality of vision areas summit, and

Described the first polygon and described the second polygon relative to each other have the back side to orientation.

88. 6 described methods according to claim 8, wherein, the described information of described at least one boundary edge of expression comprises the information that represents at least one compound-contoured summit, described at least one compound-contoured summit is the compound-contoured summit with respect to described the first vision area, described compound-contoured summit comprises and is subjected to wedge that described the first vision area supports and the cross-shaped portion between at least one single order silhouette edge, and described at least one single order silhouette edge comprises:

Share the first polygon and second polygon of described at least one single order silhouette edge,

The described first polygon back side is to each vision area summit of described the first vision area,

Described the second polygon is just being faced at least one vision area summit of described the first vision area, and

Described the first polygon and described the second polygon relative to each other have the back side to orientation.

89. 6 described methods according to claim 8, wherein, the described information that represents at least one boundary edge comprises the information of blocking boundary edge that expression first is looked from vision area, described first is by at least one wedge and the described first group of limit that the polygonal cross-shaped portion of grid forms that supported by described the first vision area from the border of blocking that vision area is looked, described at least one wedge is the wedge that is incident on the single order silhouette edge, described single order silhouette edge is the single order silhouette edge with respect to described the first vision area, and described single order silhouette edge comprises:

Share the first polygon and second polygon of described single order silhouette edge,

The described first polygon back side is to each vision area of described the first vision area,

Described the second polygon is just being faced at least one vision area summit of described the first vision area, and

Described the first polygon and described the second polygon relative to each other have the back side to orientation.

90. one kind is used institute's canned data to come to determine from the grid polygon that forms polygonal mesh to said method comprising the steps of from the first vision area first group of polygonal method of grid of looking visible:

The visible second group of grid polygon of looking from the second vision area retrieved, and described the second vision area comprises described the first vision area;

To representing that the polygonal information of at least one the first grid retrieves, described at least one first grid polygon belongs to the polygonal set of grid of connection, and as seen the polygonal described set of the grid of connection looks from described the first vision area;

Expression being blocked the information of blocking boundary edge that first in the sequence of boundary edge look from vision area retrieves, the described first boundary edge of blocking of looking from vision area is that single order with respect to described the first vision area blocks boundary edge, and described sequence comprises that the single order with respect to described the first vision area blocks the border;

Expression is retrieved with the described first information of blocking the first single order silhouette edge corresponding to boundary edge of looking from vision area;

Structure block in the described sequence of boundary edge at least one second block boundary edge from what vision area was looked, at least one second boundary edge of blocking of looking from vision area of described structure may further comprise the steps:

Identify described first from the summit of blocking boundary edge that vision area is looked, described first is formed by the first wedge and the polygonal cross-shaped portion of the second grid from the described summit of blocking boundary edge that vision area is looked, and described the first wedge is supported by described the first vision area and is incident on described the first single order silhouette edge;

Identify the 3rd grid polygon, described the 3rd grid polygon and described the first grid polygon have shared limit, and described shared limit comprises described first from the described summit of blocking the border that vision area is looked,

At described the first single order silhouette edge structure the first wedge;

Structure described at least the second blocks boundary edge from what vision area was looked by what described the first wedge and the described the 3rd polygonal cross-shaped portion formed;

Traversal belongs to polygonal described at least one set of grid of described second group of polygonal connection of grid, and described traversal comprises:

Begin described traversal from described at least the first grid polygon, and

When blocking the border, described single order interrupts described traversal when running into.

91. one kind is used institute's canned data to come to determine from the grid polygon that forms polygonal mesh to said method comprising the steps of from the first vision area first group of polygonal method of grid of looking visible:

The visible second group of grid polygon of looking from the second vision area retrieved, and described the second vision area comprises described the first vision area;

To representing that the polygonal information of at least one the first grid retrieves, described at least one first grid polygon belongs to the polygonal set of grid of connection, and as seen the polygonal described set of the grid of connection looks from described the first vision area;

Information at least one the first single order silhouette edge in the sequence of expression single order silhouette edge is retrieved, described at least one first single order silhouette edge is the single order silhouette edge with respect to described the first vision area, and described sequence comprises the single order profile contour with respect to described the first vision area;

At least another single order silhouette edge in the described sequence of identification single order silhouette edge, described at least another single order silhouette edge is connected to described at least one single order silhouette edge, and each described single order silhouette edge comprises:

Share the first polygon and second polygon of the single order silhouette edge of described correspondence,

The described first polygon back side is to each summit of described the first vision area,

Described the second polygon is just being faced at least one the vision area summit from described the first vision area, and

Described the first polygon and described the second polygon relative to each other have the back side to orientation; And

Traversal belongs to polygonal at least one set of grid of described second group of polygonal connection of grid, and described traversal comprises:

Begin described traversal from described at least one first grid polygon, and

When running into described at least one single order silhouette edge or described at least another single order silhouette edge, interrupt described traversal.

92. one kind is used institute's canned data to come to determine that described method comprises from the first vision area first group of polygonal method of grid of looking visible from the grid polygon that forms polygonal mesh:

The visible second group of grid polygon of looking from the second vision area retrieved, and described the second vision area comprises described the first vision area;

To representing that the polygonal information of at least one the first grid retrieves, described at least one first grid polygon belongs to the polygonal set of grid of connection, and as seen the polygonal described set of the grid of connection looks at least in part from described the first vision area;

Retrieve representing at least one information of blocking boundary edge, describedly block the limit that boundary edge comprises the polygonal described set of grid of connection, the described boundary edge of blocking is looked invisible from described the first vision area, represent that at least one described information of blocking boundary edge comprises limit-run into numeral, the value of described limit-run into numeral equal to run into described at least one block the required iteration number to the breadth first traversal of the polygonal described set of grid that connects of boundary edge, described breadth first traversal is from described at least one first grid polygon;

Traversal belongs to polygonal at least one set of grid of described second group of polygonal connection of grid, and described traversal comprises:

Begin breadth first traversal from described at least one first grid polygon;

After equaling described limit-run into numeral, the iteration number of described breadth first traversal interrupts described breadth first traversal.

93. one kind is used institute's canned data to determine to said method comprising the steps of from the first vision area first group of polygonal method of grid of looking visible from the grid polygon that forms polygonal mesh:

The visible second group of grid polygon of looking from the second vision area retrieved, and described the second vision area comprises described the first vision area;

To representing that the polygonal information of at least one the first grid retrieves, described at least one first grid polygon belongs to the polygonal set of grid of connection, and as seen the polygonal described set of the grid of connection looks from described the first vision area;

The information that represents at least one single order silhouette edge is retrieved, described at least one single order silhouette edge is the single order silhouette edge with respect to described the first vision area, the described information that represents at least one single order silhouette edge comprises limit-run into numeral, the value of described limit-run into numeral equals to run into the required iteration number to the breadth first traversal of the polygonal described set of grid that connects of described at least one single order silhouette edge, described breadth first traversal begins from described at least one first grid polygon, and described at least one single order silhouette edge comprises:

Share the first polygon and second polygon of described single order silhouette edge,

The described first polygon back side is to each summit of described the first vision area,

Described the second polygon is just being faced at least one the vision area summit from described the first vision area, and

Described the first polygon and described the second polygon relative to each other have the back side to orientation;

Traversal belongs to polygonal described at least one set of grid of described second group of polygonal connection of grid, and described traversal comprises:

Begin breadth first traversal from described at least one first grid polygon;

After equaling described limit-run into numeral, the iteration number of described breadth first traversal interrupts described breadth first traversal.

94. the method determining and store the information that first group of grid polygon is described, described first group of grid polygon forms polygonal mesh, and described first group of grid polygon looked invisible from the first vision area, said method comprising the steps of:

Determine and storage from the second vision area second group of grid polygon of looking visible, described the second vision area comprises described the first vision area;

Determine and the polygonal information of storage at least one grid of expression, described at least one grid polygon belongs to the polygonal set of grid of connection, and the polygonal described set of the grid of connection is looked invisible from described the first vision area;

Determine and store the information of at least one boundary edge of polygonal described at least one set of grid that expression connects.

95. one kind is used institute's canned data to determine first group of polygonal method of grid, described first group of grid polygon forms polygonal mesh, and as seen described first group of grid polygon look from the first vision area, and described method comprises:

The visible second group of grid polygon of looking from the second vision area retrieved, and described the second vision area comprises described the first vision area;

To representing that the polygonal information of at least one grid retrieves, described at least one first grid polygon belongs to the polygonal set of grid of connection, and the polygonal described set of the grid of connection is looked invisible from described the first vision area;

The information of at least one boundary edge of polygonal described at least one set of grid that expression is connected is retrieved;

Polygonal described set travels through to the grid that belongs to described second group of polygonal connection of grid, and described traversal comprises:

Begin described traversal from described at least one first grid polygon;

When running into described at least one boundary edge, interrupt described traversal.

96. 5 described methods according to claim 9, wherein, the described information that represents at least one boundary edge comprises that at least one blocks boundary edge from what vision area was looked, described at least one boundary edge of blocking of looking from vision area is the limit that is formed by at least one wedge and the polygonal cross-shaped portion of described grid, described at least one wedge is incident on the single order silhouette edge, described single order silhouette edge is the single order silhouette edge with respect to described the first vision area, and described single order silhouette edge comprises:

Share the first polygon and second polygon of described single order silhouette edge,

The described first polygon back side is to each summit of described the first vision area,

Described the second polygon is just being faced at least one the vision area summit from described the first vision area, and

Described the first polygon and described the second polygon relative to each other have the back side to orientation.

97. one kind is used institute's canned data to come to determine from the grid polygon that forms polygonal mesh to said method comprising the steps of from the first vision area first group of polygonal method of grid of looking sightless:

The visible second group of grid polygon of looking from the second vision area retrieved, and described the second vision area comprises described the first vision area;

To representing that the polygonal information of at least one the first grid retrieves, described at least one first grid polygon belongs to the polygonal set of grid of connection, and the polygonal described set of the grid of connection is looked from described the first vision area and is blocked at least in part;

Retrieve representing at least one information of exposing boundary edge, describedly expose the limit that boundary edge comprises the polygonal described set of grid of connection, as seen the described boundary edge of exposing looks from described the first vision area, represent that at least one described information of exposing boundary edge comprises limit-run into numeral, the value of described limit-run into numeral equal to run into described at least one expose the required iteration number to the breadth first traversal of the polygonal described set of grid that connects of boundary edge, described breadth first traversal is from described at least one first grid polygon;

Polygonal described at least one set travels through to the grid that belongs to described second group of polygonal connection of grid, and described traversal comprises:

Begin described breadth first traversal from described at least one first grid polygon;

After equaling described limit-run into numeral, the iteration number of described breadth first traversal interrupts described breadth first traversal.

98. a system that determines and store the information that first group of grid polygon is described, described grid polygon forms polygonal mesh, and as seen described first group of grid polygon look from the first vision area, and described system comprises:

Scrambler with processor, described processor is configured to:

Determine that described the second vision area comprises described the first vision area from the second vision area second group of grid polygon of looking visible;

Determine the polygonal information of at least one grid of expression, described at least one grid polygon belongs to the polygonal set of grid of connection, and as seen the polygonal described set of the grid of connection looks from described the first vision area; And

Determine the information of at least one boundary edge of polygonal described at least one set of grid that expression connects; And

Database, described database are stored described second group of grid polygon, the expression polygonal described information of described at least one grid and are represented the described information of at least one boundary edge.

99. 8 described methods according to claim 9, wherein, the described information that represents at least one boundary edge comprises the information that represents at least one single order silhouette edge, and described at least one single order silhouette edge is the single order silhouette edge with respect to described the first vision area, and described at least one single order silhouette edge comprises:

Share the first polygon and second polygon of described at least one single order silhouette edge,

The described first polygon back side is to each vision area summit of described the first vision area,

Described the second polygon is just being faced at least one vision area summit of described the first vision area, and

Described the first polygon and described the second polygon relative to each other have the back side to orientation.

100. 8 described systems according to claim 9, wherein, the described information that represents at least one boundary edge comprises the information that represents at least one compound-contoured summit, described at least one compound-contoured summit is the compound-contoured summit with respect to described the first vision area, described compound-contoured summit comprises and is subjected to wedge that described the first vision area supports and the cross-shaped portion between at least one single order silhouette edge, and described at least one single order silhouette edge comprises:

Share the first polygon and second polygon of described at least one single order silhouette edge,

The described first polygon back side is to each vision area summit of described the first vision area,

Described the second polygon is just being faced at least one vision area summit of described the first vision area, and

Described the first polygon and described the second polygon relative to each other have the back side to orientation.

101. 8 described systems according to claim 9, wherein, the described information that represents at least one boundary edge comprises that at least one blocks boundary edge from what vision area was looked, described at least one boundary edge of blocking of looking from vision area is the limit that is formed by at least one wedge that supported by described the first vision area and the polygonal cross-shaped portion of described grid, described at least one wedge is the wedge that is incident on the single order silhouette edge, described single order silhouette edge is the single order silhouette edge with respect to described the first vision area, and described single order silhouette edge comprises:

Share the first polygon and second polygon of described single order silhouette edge,

The described first polygon back side is to each vision area summit of described the first vision area,

Described the second polygon is just being faced at least one vision area summit of described the first vision area, and

Described the first polygon and described the second polygon relative to each other have the back side to orientation.

102. 9 described systems according to claim 9, wherein, describe the information that first group of polygonal described information of grid further comprises the length of the sequence that represents a plurality of single order silhouette edges, the described sequence of described a plurality of single order silhouette edges comprises the single order silhouette edge of at least summit-connection at least one summit that is connected to described at least one single order silhouette edge.

103. 7 described methods according to claim 9, wherein, the described information that represents at least one boundary edge comprises the polygonal information of at least one grid of expression, stops the breadth first traversal to the polygonal described set of grid that connects at described at least one grid polygon place.

104. 3 described systems according to claim 10, wherein, being illustrated in its place stops the polygonal described information of at least one grid to the breadth first traversal of the polygonal described set of grid that connects and comprises and be stored to a few limit-run into numeral, described limit-run into numeral is the iteration number that runs into the required described breadth first traversal in described at least one grid polygon limit, and the described iteration of described breadth first traversal begins from described at least one grid polygon of the polygonal described set of grid that connects.

105. one kind is used institute's canned data to determine first group of polygonal system of grid, described grid polygon forms polygonal mesh, looks as seen from the first vision area from the polygonal first group of grid polygon of described grid, and described method comprises:

Database, the storage of described database from the second vision area look visible second group of grid polygon, expression belong to connection the polygonal set of grid the polygonal information of at least one the first grid and be connected the information of at least one boundary edge of polygonal described at least one set of grid of connection; And

Demoder with processor, described processor is configured to:

Retrieval is from the second vision area second group of grid polygon of looking visible, and described the second vision area comprises described the first vision area,

The polygonal information of retrieval at least one the first grid of expression, described at least one first grid polygon belongs to the polygonal set of grid of connection, and as seen the polygonal described set of the grid of connection looks from described the first vision area,

The information of at least one boundary edge of polygonal described at least one set of grid that expression is connected is retrieved, and

Polygonal described at least one set travels through described traversal to the grid that belongs to described second group of polygonal connection of grid:

Begin from described at least one first grid polygon, and

When during described traversal, running into described at least one boundary edge, interrupt.

106. 5 described systems according to claim 10, wherein, the described information of described at least one boundary edge of expression comprises the information that represents at least one single order silhouette edge, described at least one single order silhouette edge is the single order silhouette edge with respect to described the first vision area, and described at least one single order silhouette edge comprises:

Share the first polygon and second polygon of described at least one single order silhouette edge,

The described first polygon back side is to each the vision area summit from described a plurality of vision areas summit,

Described the second polygon is just being faced at least one the vision area summit from described a plurality of vision areas summit, and

Described the first polygon and described the second polygon relative to each other have the back side to orientation.

107. 5 described systems according to claim 10, wherein, the described information of described at least one boundary edge of expression comprises the information that represents at least one compound-contoured summit, described at least one compound-contoured summit is the compound-contoured summit with respect to described the first vision area, described compound-contoured summit comprises and is subjected to wedge that described the first vision area supports and the cross-shaped portion between at least one single order silhouette edge, and described at least one single order silhouette edge comprises:

Share the first polygon and second polygon of described at least one single order silhouette edge,

The described first polygon back side is to each vision area summit of described the first vision area,

Described the second polygon is just being faced at least one vision area summit of described the first vision area, and

Described the first polygon and described the second polygon relative to each other have the back side to orientation.

108. 5 described systems according to claim 10, wherein, the described information that represents at least one boundary edge comprises the information of blocking boundary edge that expression first is looked from vision area, the described first boundary edge of blocking of looking from vision area is by at least one wedge and the described first group of limit that the polygonal cross-shaped portion of grid forms that supported by described the first vision area, described at least one wedge is the wedge that is incident on the single order silhouette edge, described single order silhouette edge is the single order silhouette edge with respect to described the first vision area, and described single order silhouette edge comprises:

Share the first polygon and second polygon of described single order silhouette edge,

The described first polygon back side is to each vision area of described the first vision area,

Described the second polygon is just being faced at least one vision area summit of described the first vision area, and

Described the first polygon and described the second polygon relative to each other have the back side to orientation.

109. non-transient state computer-readable recording medium with executable instruction stored thereon, when carrying out described executable instruction by processor, described executable instruction is so that the method for the information that first group of grid polygon is described is determined and stored to described processor, described grid polygon forms polygonal mesh, as seen described first group of grid polygon look from the first vision area, said method comprising the steps of:

Determine and storage from the second vision area second group of grid polygon of looking visible, described the second vision area comprises described the first vision area;

Determine and the polygonal information of storage at least one grid of expression, described at least one grid polygon belongs to the polygonal set of grid of connection, and as seen the polygonal described set of the grid of connection looks from described the first vision area; And

Determine and store the information of at least one boundary edge of polygonal at least one set of grid that expression connects.

110. 9 described non-transient state computer-readable recording mediums according to claim 10, wherein, the described information that represents at least one boundary edge comprises the information that represents at least one single order silhouette edge, described at least one single order silhouette edge is the single order silhouette edge with respect to described the first vision area, and described at least one single order silhouette edge comprises:

Share the first polygon and second polygon of described at least one single order silhouette edge,

The described first polygon back side is to each vision area summit of described the first vision area,

Described the second polygon is just being faced at least one vision area summit of described the first vision area, and

Described the first polygon and described the second polygon relative to each other have the back side to orientation.

111. 9 described non-transient state computer-readable recording mediums according to claim 10, wherein, the described information that represents at least one boundary edge comprises the information that represents at least one compound-contoured summit, described at least one compound-contoured summit is the compound-contoured summit with respect to described the first vision area, described compound-contoured summit comprises and is subjected to wedge that described the first vision area supports and the cross-shaped portion between at least one single order silhouette edge, and described at least one single order silhouette edge comprises:

Share the first polygon and second polygon of described at least one single order silhouette edge,

The described first polygon back side is to each vision area summit of described the first vision area,

Described the second polygon is just being faced at least one vision area summit of described the first vision area, and

Described the first polygon and described the second polygon relative to each other have the back side to orientation.

112. 9 described non-transient state computer-readable recording mediums according to claim 10, wherein, the described information that represents at least one boundary edge comprises that at least one blocks boundary edge from what vision area was looked, described at least one boundary edge of blocking of looking from vision area is the limit that is formed by at least one wedge that supported by described the first vision area and the polygonal cross-shaped portion of described grid, described at least one wedge is the wedge that is incident on the single order silhouette edge, described single order silhouette edge is the single order silhouette edge with respect to described the first vision area, and described single order silhouette edge comprises:

Share the first polygon and second polygon of described single order silhouette edge,

The described first polygon back side is to each vision area summit of described the first vision area,

Described the second polygon is just being faced at least one vision area summit of described the first vision area, and

Described the first polygon and described the second polygon relative to each other have the back side to orientation.

113. 0 described non-transient state computer-readable recording medium according to claim 11, wherein, describe the information that first group of polygonal described information of grid further comprises the length of the sequence that represents a plurality of single order silhouette edges, the described sequence of described a plurality of single order silhouette edges comprises the single order silhouette edge of at least summit-connection at least one summit that is connected to described at least one single order silhouette edge.

114. 9 described non-transient state computer-readable recording mediums according to claim 10, wherein, the described information that represents at least one boundary edge comprises the information that represents at least one grid polygon limit, stops the breadth first traversal to the polygonal described set of grid that connects at described at least one place, grid polygon limit.

115. 4 described non-transient state computer-readable recording mediums according to claim 11, wherein, being illustrated in its place stops the polygonal described information of at least one grid to the breadth first traversal of the polygonal described set of grid that connects and comprises and be stored to a few limit-run into numeral, described limit-run into numeral is the iteration number that runs into the required described breadth first traversal of described at least one grid polygon, and the described iteration of described breadth first traversal begins from described at least one grid polygon of the polygonal described set of grid that connects.

116. one kind is used institute's canned data to determine first group of polygonal method of grid, described grid polygon forms polygonal mesh, looks as seen from the first vision area from the polygonal first group of grid polygon of described grid, and described method comprises:

The visible second group of grid polygon of looking from the second vision area retrieved, and described the second vision area comprises described the first vision area;

To representing that the polygonal information of at least one the first grid retrieves, described at least one first grid polygon belongs to the polygonal set of grid of connection, and as seen the polygonal described set of the grid of connection looks from described the first vision area;

The information of at least one boundary edge of polygonal described at least one set of grid that expression is connected is retrieved;

Polygonal described at least one set travels through to the grid that belongs to described second group of polygonal connection of grid, and described traversal comprises:

Begin described traversal from described at least one first grid polygon, and

When during described traversal, running into described at least one boundary edge, interrupt described traversal.

117. 6 described non-transient state computer-readable recording mediums according to claim 11, wherein, the described information of described at least one boundary edge of expression comprises the information that represents at least one single order silhouette edge, described at least one single order silhouette edge is the single order silhouette edge with respect to described the first vision area, and described at least one single order silhouette edge comprises:

Share the first polygon and second polygon of described at least one single order silhouette edge,

The described first polygon back side is to each the vision area summit from described a plurality of vision areas summit,

Described the second polygon is just being faced at least one the vision area summit from described a plurality of vision areas summit, and

Described the first polygon and described the second polygon relative to each other have the back side to orientation.

118. 7 described non-transient state computer-readable recording mediums according to claim 11, wherein, the described information of described at least one boundary edge of expression comprises the information that represents at least one compound-contoured summit, described at least one compound-contoured summit is the compound-contoured summit with respect to described the first vision area, described compound-contoured summit comprises and is subjected to wedge that described the first vision area supports and the cross-shaped portion between described at least one single order silhouette edge, and described at least one single order silhouette edge comprises:

Share the first polygon and second polygon of described at least one single order silhouette edge,

The described first polygon back side is to each vision area summit of described the first vision area,

Described the second polygon is just being faced at least one vision area summit of described the first vision area, and

Described the first polygon and described the second polygon relative to each other have the back side to orientation.

119. 6 described non-transient state computer-readable recording mediums according to claim 11, wherein, the described information that represents at least one boundary edge comprises the information of blocking boundary edge that expression first is looked from vision area, the described first boundary edge of blocking of looking from vision area is by at least one wedge and the described first group of limit that the polygonal cross-shaped portion of grid forms that supported by described the first vision area, described at least one wedge is the wedge that is incident on the single order silhouette edge, described single order silhouette edge is the single order silhouette edge with respect to described the first vision area, and described single order silhouette edge comprises:

Share the first polygon and second polygon of described single order silhouette edge,

The described first polygon back side is to each vision area of described the first vision area,

Described the second polygon is just being faced at least one vision area summit of described the first vision area, and

Described the first polygon and described the second polygon relative to each other have the back side to orientation.

120. one kind is used for the control graphical information to the method based on the transmission of looking ahead of navigation of client device what server was carried out, described graphical information comprises looks visible and from the first vision area the first block graphics information of looking sightless from the second vision area, and described graphical information comprises looks visible and from described the first vision area the second block graphics information of looking sightless from described the second vision area, the level of detail of described the second block graphics information is lower than the level of detail of described the first block graphics information, said method comprising the steps of:

A) determine to be positioned at very first time section during described the first block graphics information will arrive after described the first block graphics information on the described client device in the described client device access of scheduling; And

B) during described very first time section, send described the second block graphics information.

121. 0 described method according to claim 12 wherein, describedly may further comprise the steps based on looking ahead of navigation:

Receive the information of expression first the first viewpoint position constantly from described client device;

Predict second the second viewpoint position of constantly locating with the described information that represents described the first viewpoint position, described second is later than described first constantly constantly;

Determine whether described the first viewpoint position is positioned at different vision areas with described the second viewpoint position,

When having determined that described the first viewpoint position is positioned at different vision areas with described the second viewpoint position, send described the second block graphics information.

122. 0 described method according to claim 12, wherein, described graphical information comprises polygon.

123. 0 described method according to claim 12 is further comprising the steps:

C) send additional graphical information during the second time period that occurs after described very first time section, described additional graphical information comprises for the information that details is added into described the second block graphics information that sends during described very first time section.

124. 3 described methods according to claim 12, wherein, according to the amount that is chosen in the described additional patterns information that sends in the described step c) at described very first time section and the function that passes the time period between described the second time period.

125. 4 described methods according to claim 12, wherein, with the human visual system described pass the time period after the ability of the described additional patterns information of perception further be chosen in pro rata the amount of the described additional patterns information that sends in the described step c).

126. 3 described methods according to claim 12 wherein, further are chosen in the amount of the described additional patterns information that sends in the described step c) pro rata with the described additional patterns information of human visual system's perception as the ability of exposing the function of time.

127. method of looking ahead and sending based on navigation for the graphical information that controls to client device of carrying out at server, described graphical information comprises looks visible and from the first vision area the first block graphics information of looking sightless from the second vision area, and described graphical information comprises looks visible and from described the first vision area the second block graphics information of looking sightless from described the second vision area, the level of detail of described the second block graphics information is lower than the level of detail of described the first block graphics information, said method comprising the steps of:

A) determine during very first time section the first block graphics information sent to the required first data transmission speed of described client unit;

B) determine in the maximum available data transmission speed between described server and the described client device during the described very first time section; And

C) when having determined described first data transmission speed greater than described maximum available data transmission speed, during described very first time section, described the second block graphics information is sent to described client device.

128. method of looking ahead and sending based on navigation for the graphical information that controls to client device of carrying out at server, described graphical information comprises looks visible and from the first vision area the first block graphics information of looking sightless from the second vision area, and described graphical information comprises looks visible and from described the first vision area the second block graphics information of looking sightless from described the second vision area, the level of detail of described the second block graphics information is lower than the level of detail of described the first block graphics information, said method comprising the steps of:

A) image-space velocity of determining at least one graphic element of being shown by described client device surpasses the very first time section during the predetermined value;

B) when the described image-space velocity of described at least one graphic element of having determined to be shown by described client surpasses predetermined value, during described very first time section, described the second block graphics information is sent to described client device.

129. method of looking ahead and sending based on navigation for the graphical information that controls to client device of carrying out at server, described graphical information comprises looks visible and from the first vision area the first block graphics information of looking sightless from the second vision area, and described graphical information comprises looks visible and from described the first vision area the second block graphics information of looking sightless from described the second vision area, the level of detail of described the second block graphics information is lower than the level of detail of described the first block graphics information, said method comprising the steps of:

A) the estimation retina speed of determining at least one graphic element of being shown by described client device surpasses the very first time section during the predetermined threshold;

B) when the described estimation retina speed of described at least one graphic element of having determined to be shown by described client device surpasses predetermined threshold, during described very first time section, described the second block graphics information is sent to client device.

130. method of looking ahead and sending based on navigation for the graphical information that controls to client device of carrying out at server, described graphical information comprises looks visible and from the first vision area the first block graphics information of looking sightless from the second vision area, and described graphical information comprises looks visible and from described the first vision area the second block graphics information of looking sightless from described the second vision area, the level of detail of described the second block graphics information is lower than the level of detail of described the first block graphics information, said method comprising the steps of:

A) image-space velocity of determining at least one graphic element of being shown by described client device has surpassed the very first time section during the predetermined value, and described at least one graphic element is illustrated in described the first block graphics information and described the second block graphics information;

B) when the described image-space velocity of described at least one graphic element of having determined to be shown by described client device has surpassed described predetermined value, during described very first time section, described the second block graphics information is sent to described client device.

131. method of looking ahead and sending based on navigation for the graphical information that controls to client device of carrying out at server, described graphical information comprises looks visible and from the first vision area the first block graphics information of looking sightless from the second vision area, and described graphical information comprises looks visible and from described the first vision area the second block graphics information of looking sightless from described the second vision area, the level of detail of described the second block graphics information is lower than the level of detail of described the first block graphics information, said method comprising the steps of:

A) the estimation retina speed of determining at least one graphic element of being shown by described client device surpasses the very first time section during the predetermined value, and described at least one graphic element is illustrated in described the first block graphics information and described the second block graphics information;

B) when the described estimation retina speed of described at least one graphic element of having determined to be shown by described client device surpasses described predetermined value, during described very first time section, described the second block graphics information is sent to described client device.

132. a control to look visible from the second vision area from the look method of determining of sightless the first block graphics information of the first vision area, described graphical information comprises at least one first polygonal mesh, described the second vision area is adjacent with described the first vision area, said method comprising the steps of:

A) definite size of uniting the vision area that forms by described the first vision area and described the second vision area;

B) determine distance between the described associating of described at least one first polygonal mesh and described the first vision area and described the second vision area;

C) determine the ratio of described distance and described size; And

D) when having determined described ratio less than predetermined value, determine to look visible and from described the first vision area the first block graphics information of looking sightless from described the second vision area.

133. a control to look visible from the second vision area from the look method of determining of sightless the first block graphics information of the first vision area, described graphical information comprises at least one first polygonal mesh, described the second vision area is adjacent with described the first vision area, said method comprising the steps of:

A) definite size of uniting the vision area that forms by described the first vision area and described the second vision area;

B) determine distance between the described associating of described at least one first polygonal mesh and described the first vision area and described the second vision area;

C) determine the ratio of described distance and described size; And

D) when having determined described ratio greater than predetermined value, determine described the 3rd block graphics information of looking visible of uniting from described the first vision area and described the second vision area.

134. method of looking ahead and sending based on navigation that is used for controlling to client device of carrying out at server, described graphical information comprises from the second vision area looks visible and from the first vision area the first block graphics information of looking sightless, said method comprising the steps of:

A) determine to be positioned at very first time section during described the first block graphics information will arrive after described the first block graphics information on the described client device in the described client device access of scheduling; And

B) send adjustment information during described very first time section, described adjustment information is used for reducing viewpoint speed.

135. 4 method according to claim 13 wherein, describedly may further comprise the steps based on looking ahead of navigation:

Receive the information of expression first the first viewpoint position constantly from described client device;

The described information prediction second of described the first viewpoint position of use expression is second viewpoint position at place constantly, and described second is later than described first moment constantly;

Determine whether described the first viewpoint position is positioned at different vision areas with described the second viewpoint position,

When having determined that described the first viewpoint position is positioned at different vision areas with described the second viewpoint position, send described the second block graphics information.

136. one kind is used for control from the method for looking ahead and receiving based on navigation of the graphical information of server apparatus what client device was carried out, described graphical information comprises from the second vision area looks visible and from the first vision area the first block graphics information of looking sightless, said method comprising the steps of:

A) determine to be positioned at very first time section during described the first block graphics information will arrive after the above first block graphics information of described client device in the described client device access of scheduling; And

B) accept adjustment information during described very first time section, described adjustment information is used for reducing viewpoint speed.

137. 6 described methods according to claim 13 wherein, are describedly carried out on the described server apparatus and further comprising the steps based on being taken in advance of navigation:

Receive the information of expression first the first viewpoint position constantly from described client device,

The described information prediction second of described the first viewpoint position of use expression is second viewpoint position at place constantly, and described second is later than described first moment constantly;

Determine whether described the first viewpoint position is positioned at different vision areas with described the second viewpoint position, and

When having determined that described the first viewpoint position is positioned at different vision areas with described the second viewpoint position, send described the second block graphics information.

138. 7 described methods according to claim 13 are further comprising the steps:

C) during described very first time section, show the expression corresponding with the described viewpoint speed that has reduced.

139. 8 described methods according to claim 13, wherein, the described expression corresponding with the described viewpoint speed that has reduced comprises the expression of physics fatigue.

140. 8 described methods according to claim 13, wherein, the described expression corresponding with the described viewpoint speed that has reduced comprises the expression of engine fatigue.

141. 8 described methods according to claim 13, wherein, the described expression corresponding with the described viewpoint speed that has reduced comprises the expression of insufficient engine power.

142. system of looking ahead and sending based on navigation that controls graphical information, described graphical information comprises looks visible and from the first vision area the first block graphics information of looking sightless from the second vision area, and described graphical information comprises looks visible and from described the first vision area the second block graphics information of looking sightless from described the second vision area, the level of detail of described the second block graphics information is lower than the level of detail of described the first block graphics information, and described system comprises:

Server with processor, described processor is configured to:

A) determine to be positioned at very first time section during described the first block graphics information will arrive after described the first block graphics information on the described client device in the described client device access of scheduling, and

B) during described very first time section, send described the second block graphics information; And client device, described client device shows described the second block graphics.

143. 2 described systems according to claim 14, wherein, the described processor of described server also is configured to:

Receive the information of expression first the first viewpoint position constantly from described client device,

The described information prediction second of described the first viewpoint position of use expression is second viewpoint position at place constantly, and described second is later than described first moment constantly;

Determine whether described the first viewpoint position is positioned at different vision areas with described the second viewpoint position,

When having determined that described the first viewpoint position is positioned at different vision areas with described the second viewpoint position, send described the second block graphics information.

144. 2 described systems according to claim 14, wherein, described graphical information comprises polygon.

145. 2 described systems according to claim 14, wherein, the described processor of described server further is configured to:

C) send additional graphical information during the second time period that occurs after described very first time section, described additional graphical information comprises for the information that details is added into described the second block graphics information that sends in described very first time section.

146. 5 described systems according to claim 14 wherein, are chosen in the amount of the described additional graphical information that sends in the described step c) according to described very first time section and the function that passes the time period between described the second time period.

147. 6 described systems according to claim 14, wherein, with the human visual system further be chosen in pro rata the amount of the described additional graphical information that sends in the described step c) in the described ability that passes the described additional graphical information of perception after the time period.

148. 5 described systems according to claim 14, wherein, the ability of exposing the described additional graphical information of the perception of function of time with human visual system's conduct is chosen in the amount of the described additional graphical information that sends in the described step c) pro rata.

149. non-transient state computer-readable recording medium with executable instruction stored thereon, when carrying out described executable instruction by the processor in the server, described executable instruction is so that described processor carries out the method for looking ahead and sending based on navigation be used to the graphical information that controls to client device, described graphical information comprises looks visible and from the first vision area the first block graphics information of looking sightless from the second vision area, and described graphical information comprises looks visible and from described the first vision area the second block graphics information of looking sightless from described the second vision area, the level of detail of described the second block graphics information is lower than the level of detail of described the first block graphics information, said method comprising the steps of:

A) determine to be positioned at very first time section during described the first block graphics information will arrive after described the first block graphics information on the described client device in the described client device access of scheduling;

B) during described very first time section, send described the second block graphics information.

150. 9 described non-transient state computer-readable recording mediums according to claim 14, wherein, described executable instruction is also so that described processor carries out following method step:

Receive the information of expression first the first viewpoint position constantly from described client device;

The described information prediction second of described the first viewpoint position of use expression is second viewpoint position at place constantly, and described second is later than described first moment constantly;

Determine whether described the first viewpoint position is positioned at different vision areas with described the second viewpoint position,

When having determined that described the first viewpoint position is positioned at different vision areas with described the second viewpoint position, send described the second block graphics information.

151. 8 described non-transient state computer-readable recording mediums according to claim 14, wherein, described graphical information comprises polygon.

152. 9 described non-transient state computer-readable recording mediums according to claim 14, wherein, described executable instruction is further so that described processor carries out following method step:

C) send additional graphical information during the second time period that occurs after described very first time section, described additional graphical information comprises for the information of details being added to described the second block graphics information that sends during described very first time section.

153. 2 described non-transient state computer-readable recording mediums according to claim 15, wherein, be chosen in the amount of the described additional graphical information that sends in the described step c) according to described very first time section and the function that passes the time period between described the second time period.

154. 3 described non-transient state computer-readable recording mediums according to claim 15, wherein, with the human visual system further be chosen in pro rata the amount of the described additional graphical information that sends in the described step c) in the described ability that passes the described additional graphical information of perception after the time period.

155. 2 described non-transient state computer-readable recording mediums according to claim 15, wherein, be chosen in pro rata the amount of the described additional graphical information that sends in the described step c) as the ability of exposing the function of time with the described additional graphical information of human visual system's perception.

Images